Tweet

by Drew Rosielle (@drosielle)

A Satirical Monologue in One Act:

“Ok, 3rd year resident, let’s talk about rotating opioids. What do I mean by ‘rotating’ opioids? It’s just therapeutically switching one opioid with another. It’s um, like, a turnstile, I guess? Anyway--first let’s look at this equianalgesic table. Do you know what equianalgesia means? No? It is the concept that different opioids have the same analgesic power but at different milligram doses due to different potencies. That is, the idea that, say, 50 mg of oral morphine has the same analgesic power as 10 mg of hydromorphone. So 50 mg of oral morphine is equivalent to 10 mg of oral hydromorphone.

"Ok, let’s look at this table, every entry on this grid is equianalgesic, meaning that the 30 mg here of oral morphine has the same analgesic power as the 1.5 mg of IV hydromorphone. So, let’s say we have a patient on 100 mg oral morphine. How much hydromorphone is that “equivalent” to? Ok, so what we have to do is set up a cross multiplication problem…here’s how you do that.

[We’ll skip the painful talk-through of the cross multiplication.]

"Ok, so it looks like the answer is 5 mg of hydromorphone IV. Now we have to adjust that dose for incomplete cross-tolerance. Do you know what that is? No? Ok, so incomplete-cross tolerance is the idea that our patient on 100 mg of morphine is somewhat tolerant to the morphine. They are therefore also presumably somewhat tolerant to any other opioid agonist like hydromorphone. The problem is that the tolerance they have to morphine may not fully apply to the tolerance they have to the hydromorphone—the morphine tolerance “incompletely” transfers so to speak to the patient’s hydromorphone tolerance. So, what that means is that even though this table says 100 mg of PO morphine is equal to 5 mg IV hydromorphone, for safety’s sake we need to reduce that 5 mg somewhat to determine what we actually put the patient on. That’s what we mean by reducing for incomplete cross-tolerance.

"If you have 5 minutes, it’s actually an interesting story, because no one is really sure that ‘incomplete cross-tolerance’ is even physiologically a phenomenon, like on the cellular level, our field just calls it that because that’s what it’s always been called. In fact, the idea of ‘incomplete cross-tolerance’ only makes sense if you think there actually is a firm “equianaglesic” potency relationship between morphine and hydromorphone that applies to most patients. In fact, no one actually believes that and the broad consensus is that there is a pretty significant range of ‘equianalgesia’ between any 2 opioids in different patients. I.e., in one patient, 100 mg of oral morphine may end up providing equal analgesia to 12 mg IV hydromorphone, in another patient, 4 mg of IV hydromorphone, etc. It’s a little bit of a crap-shoot.

"Actually, it’s probably far worse even with fentanyl, we’re talking about a far, far wider range of individual responses, and don’t even get me started talking about methadone and buprenorphine [self-knowing chuckle here]!

"So, back to incomplete cross-tolerance. If you’re going to argue that it would just be easier to say ‘dose reductions due to safety’ or ‘dose reductions due to individual variability in responding to opioids’, you’d be right. That’s much easier to understand, and probably more accurate, but maybe it makes us feel like we’re the keepers on secret knowledge to say incomplete cross-tolerance? Who knows?

"Regardless, this is where it gets really interesting, because to properly account for ‘incomplete cross-tolerance’ we have to decide how much less opioid we need to put someone on after we do the initial calculation. Yes, there are 2 calculations: the equianalgesic one, then a 2nd incomplete cross-tolerance one in which we reduce the first amount by a further amount, most people recommend 25-50%.

"How do you choose how much you actually dose reduce? Well, there’s about 10 patient factors you need to balance when making that decision. Yes, I said 10, but don’t worry, if you do this for a couple years it’ll become second nature to you [self-knowing chuckle]!

"Ok, so let’s talk about these 10 patient factors….”

[End]

Dear friends and colleagues, I think we can all do better than this, and here is my current thinking as to a better approach.

In my last post I outlined why I think equianalgesic tables (EATs) are broken and proposed we collectively move instead to using Conversion Tables (CTs) when teaching others how to switch opioids. Briefly, this is because CTs engender much simpler and clearer math; and we can easily adjust the conversion factors in CTs based on emerging data, without having to rejigger every ‘equianalgesic’ relationship which EATs force us to do. My offer from the last blog post still stands: it would be ideal if there could be a consensus process to generate a consensus CT that our professional community can broadly adopt. While I like the CT I made, I don’t think any one person should be setting the conversion ratios alone, even if it is standing on the shoulders of giants! I’m not an investigator but am willing to lend any support I can to this. It is also possible the best option is for organized medicine to lead this, and I’ve been chatting with some folks about this a little already. I’m not sure where it will lead, nor if AAHPM has any appetite for that, we’ll see.

Regardless, I’ve been thinking and talking with others a lot about the entire enterprise of how we teach and talk about opioid switching and I have a lot of ideas that go beyond CTs for potential ways we can improve our approach to this. This post is simply to capture those ideas, none of which I’m claiming are independently mine.

1) We should never utter the phrase “incomplete cross-tolerance” again.

I hope this is not controversial, I’ve yet to meet anyone in the “Incomplete Cross-Tolerance: What a Super Helpful Concept!” camp, but its use continues to pervade discussions of opioid switching. My fundamental objections to incomplete cross-tolerance: it is a concept that requires an entire paragraph of explanation (!!) to someone unfamiliar with it, plus it may not even be a ‘real’ thing, plus there are readily available, broadly understood, completely accurate alternatives to it (e.g., dose reductions for safety, dose reductions due to individual variability, etc.). While the monologue that started this post is satire, I’m sure I’ve uttered every phrase in it for real at one point or the other in the past 20 years. I know I am not alone.

2) We should talk about ‘methods for safe opioid switching’ and not equianalgesia. Equianalgesia is unnecessarily jargony and I don’t think it even exists on the population level.

Incomplete cross-tolerance only makes sense if we think equianalgesia itself is a firm and helpful concept. If you believe that we can announce, without qualification, that 20 mg of IV hydromorphone is ‘equianalgesic’ to 100 mg of oral morphine then maybe it makes sense to talk about incomplete cross-tolerance. But I don’t know of anyone who thinks that “100 mg of oral morphine is equivalent to 20 mg of IV hydromorphone” in any sort of widely generalizable, arbitrary way. I think all of us operating in this realm may acknowledge that across 100 patients maybe the averages end up close to 100:20, but who cares? Honestly, that factoid is a DISTRACTION because the way EAT switching has currently been set up has encouraged us to first focus on this fake-and-based-in-less-science-than-we’d-want notion of equianalgesia, then prompted us to do this secondary work-around-caused-only-by-the-fact-that-we-set-a-ratio-in-the-first place. Why do it this way? Why not just switch at a ratio that works most of the time, instead of nearly none of the time?

The clinical task we are faced with is not an abstract exercise in the variable potencies of different opioid molecules averaged across the human population, instead it’s a practical task of How do I safely switch this patient’s opioids? Safely means two things here: that the patient does not have significant sedation from the switch nor a major worsening of their pain. I do think what I’m arguing here is more than semantics—that going from equianalgesia to ‘methods of safe switching’ is not just me renaming it. It is a significant conceptual frame shift in this critical and common.

Using a concept like ‘equianalgesia’ promotes this idea that there’s a “right answer” to the question of “If I’m switching someone from 100 mg of morphine in a day to IV hydromorphone how much hydromorphone do I prescribe?”

My proposal is there isn’t a right answer. It is not a helpful goal or concept and we need to move onto the more salient task of switching safely. That framework focuses us on a practical clinical task. From talking to colleagues, many of us who prescribe methadone and buprenorphine long ago moved on from any idea that it was our task to somehow divine the ‘equianalgesic’ dose of methadone from another opioid. Instead, we have a protocol to make the switch, the protocol works (‘is safe’), all is well with the world. My proposal here is that this is how we conceptualize all opioid conversions.

3) We must start emphasizing that the absolute dose of opioids involved matters for safe switching.

Stepping back from the finer points of what conversion ratios we should be using for opioids—honestly, if what you’re talking about is doses in the range, say, of 40 mg of oral morphine, I don’t really think it matters what ratios you use to switch. Frankly, it probably doesn’t matter much if you reduce further for safety or not.

Consider someone is on 40 mg a day of morphine PO. You want to put them on IV hydromorphone:

-- Classic EAT: divide by 20 to get 2 mg of IV hydromorphone

-- DOC2 Table: divide by 12.5 to get 3.2 mg of IV hydromorphone

If you reduce further for safety, presumably you’ll still get an answer about 1.2 mg apart between the 2 tables. Honestly, I don’t think the difference there is much of a safety concern. Both are fine. No opioid- tolerant patient became sedated or had a pain crisis because of a difference of 1.2 mg of IV hydromorphone over 24 hours; of course, the ‘fine-tuning’ might be a bit different, but as long as a patient had close follow-up the analgesic outcomes would be about the same too.

But what if the patient was on 400 mg a day of oral morphine? Personally, I would not be so cavalier about saying there is no relevant difference between 20 and 32 mg of IV hydromorphone in a day.

Yet, if you read pretty much all the opioid conversion stuff out there, they all treat these opioid switching ratios as fixed across all ranges of opioid doses. They suggest the same approach for 400 mg of morphine as 40 mg of morphine. To me, this is shocking, and profoundly unsafe. I’ve never met a single person who treats opioid switching ratios as linear across all opioid doses, meaning they would be more ‘conservative’ in their switching for someone going from 400 mg of morphine than 40 mg. I believe I am merely naming here a clinical practice which is already widespread! Yet, for the last twenty years of journal articles, books, national presentations, and local presentations that I’m aware of, this is not splattered in flashing neon ALL CAPS at the top of every table. It is not a core part of our basic nor specialty teaching on this topic. I genuinely don’t understand it.

Fundamentally, I’m arguing that we’ve heaped HUGE amounts of attention on the complex subtleties and nuances of reducing for safety when it comes to opioid switching, and virtually no attention to the nuances of switching at higher dose ranges. I think we got this wrong.

All this leads me to a few practical proposals:

1) “We”* stop using EATs and instead use simple “going from X to Y” conversion tables (see this post for more details).

2) We frame this practice as methods of ‘safe switching’ and do away with ‘equianalgesia’ (at least as a core concept in teaching others about opioid switching) and ‘incomplete cross-tolerance’ (that we can discard completely).

3) We promulgate a conversion table without need for additional dose reductions in most patient scenarios. I appreciate this is in some ways the most radical thing I’m proposing here, and certainly it is the one I’m most tentative about. But I suspect that a lot of our need for “reducing for incomplete cross-tolerance” has come from the fact that in order for any of our EATs to make sense they’ve needed to have on them conversion ratios that are dangerously aggressive. E.g., the for the Classic EAT I would divide the PO morphine dose by 20 to get the IV hydromorphone dose and, truth be told, it would be rare I would dose reduce beyond that: it was a good safe ratio much of the time in my experience. However, the reverse conversion, multiplying the IV hydromorphone dose by 20 to get the PO morphine dose, I regularly, heavily dose reduced (often by 50%). My argument is that to a large extent our need to create that 2nd step of dose-reduction really comes from the fact that we locked ourselves into aggressive conversion ratios with EATs and if we had just a simple “Going from X to Y” conversion table with pre-set ‘safe’/conversative ratios, we can side step this incredibly complex beast we’ve created (the need to reduce for safety / “incomplete cross tolerance”).

How would we do this: my idea is that we create a conversion table which folks broadly agree they’d be comfortable handing to a first month medical intern and telling them: “Young doctor, within these parameters, just use this.” If this idea freaks you out, remember for decades we’ve handed them inscrutable equianalgesic tables with even more inscrutable and hand-waving suggestions about reducing for incomplete cross-tolerance and expected them to know what to do. My hypothesis here is that the method I’m proposing is safer and requires far less needless work for our generalist colleagues.

As an example, we’d create a conversion table for a hypothetical 70-year-old with an eGFR of 50 and mild frailty—something like that. The idea being that most of the conversions happening out there are for patients like that or more physiologically robust, so the table would be built-in safety for a broad range of patients. I think it could be close to my draft Conversion table (Dropbox, Google Drive), but some of the conversions would be more conservative for sure.

The table would need to be explicitly labeled with the dose limits (E.g., this is for patients on 100 mg or less a day of oral morphine, 20 mg a day or less of IV hydromorphone, etc. This would avoid unnecessary calculations of oral morphine equivalents (OMED, OMDD, OMD, or any variation on those abbreviations). What those ‘dose caps’ are would require, I imagine, vigorous discussion, and I don’t strongly believe the, e.g., 100 mg a day morphine cap is actually the right one.

The table and associated teaching strongly emphasize the structure of care and close follow up is what’s most important for switching opioids safely. Don’t switch and walk away, switch and follow up!

Importantly, I’m proposing this is a conversion table for ‘generalists’ who switch opioids. I think the role of palliative and pain specialists is to have a deep knowledge of all this, including the history of opioid switching, and the nuances of dose reductions for safety, etc. All our fellows (and all our colleagues) should read Demystifying Opioid Conversions 2 Ed (aka DOC2) if they haven’t already! However, that book is not for generalists, they aren’t reading it and we can’t expect them to read it. To be clear—I’m not suggesting we jettison that book or the accumulated knowledge of opioids switching, I’m more proposing there are simpler and arguably a little safer way to teach our generalist colleagues about this, keeping in mind that to date we’ve made it so difficult for them most of them are using online calculators of various levels of reliability!

This is just a proposal which has come out of one person’s brain. I very much believe that ideally this is best decided by consensus and committee (in the absence of good clinical data for which there are zero). So, it is meant as a starting point for discussion. Ideally, I’m hoping others will join me and I am talking with CAPC about this and plan on with AAHPM too. I think the new data from MD Anderson which lead to the change in the DOC2 Table are a great opportunity for us as a field to collectively look at all this and try to do something that is far simpler, and at least a little safer. If you’re interested, let me know!

(This post was updated May 3, 2022 to correct an error in the opioid calculations!)

For more Pallimed posts about opioids.
For more Pallimed posts by Dr. Rosielle click here.


Drew Rosielle, MD is a palliative care physician at the University of Minnesota M Health Fairview in Minneapolis. He founded Pallimed in 2005. You can occasionally find him on Twitter at @drosielle.

Tweet

by Drew Rosielle (@drosielle)

I am proposing we do away with equianalgesic table (EAT) as a tool to inform clinical decisions about opioid rotations/conversions. Fundamentally, EATs create too many problems, and there are simpler and safer ways to teach clinicians how to convert between different opioids.

Part 1: New Data Can't Fix the EAT

A couple HPM fellows every year ask me which table do I prefer to use—the old EAT or the new one? By the old one, they refer to the table most of us used or were at least deeply familiar with for the last 10-20 years. By the new one, they mean the one created by Dr. Mary Lynn McPherson, PharmD in her landmark book Demystifying Opioid Conversions, 2nd Ed. If you haven’t read the book, please do, it’s really one of the best things ever written about opioid conversions. My disagreement with the book, which is admittedly a big one, is that the book promulgates the use of EATs, and my entire argument today is really in some ways inspired by the conversations I’ve had with my fellows and others as we grapple with the new EAT in her book. (From here, I’ll refer to the tables as ‘Classic’ and ‘DOC2’.)

Fundamentally, my argument is EATs themselves are intrinsically flawed: they ‘force errors’ in a way that’s entirely unnecessary and avoidable if one just doesn’t use EATs as clinical decision aides (I’ll elaborate on that statement below). Dr. McPherson’s curating and teaching new, reliable clinical data is critically important, yet better data in a flawed model like the EAT creates as many problems as it solves. I feel a little awkward putting all this out like this. Dr. McPherson is one of my heroes. The DOC2 EAT crystallized concerns I’d had over the years, and the followinghas come from conversations about EATs prompted by me and my colleagues grappling with a new table. The new DOC2 EAT is clearly better in important ways, but I also think recapitulates all the problems intrinsic to EATs, so much so we collectively should consider doing away with them.

The 2010 table is the Classic EAT, the 2018 table is the DOC2 EAT. You can see it’s changed throughout, but the big changes are the around hydromorphone, with the biggest changechanging the20:1 PO morphine:IV hydromorphone ratio to a 12.5:1 ratio.



The big changes in the DOC2 EAT are based on this study by Dr. Akhila Reddy and colleagues looking at real life data about converting hospitalized cancer patients from IV hydromorphone to PO morphine, PO hydromorphone, or PO oxycodone. Reddy et al found that the average IV HM to other drug ratios were 1:2.5 for oral hydromorphone, 1:11 for oral morphine, and 1:8 for oral oxycodone (meaning in this population, a patient on 10 mg/24 hours of IV HM, on average would end up on 25 mg PO HM / 110 mg PO MS / 80 mg PO oxycodone per 24 hours). Importantly these doses are lower than what the Classic EAT would predict, and for PO morphine and hydromorphone, a lot lower (Classic EAT promotes a 1:5 ratio for IV HM:PO HM; 1:20 ratio for IV HM:PO MS; and 1:10 ratio for IV HM:PO oxycodone)!

To the best of my knowledge, the DOC2 EAT exists largely to better account for these ratios going from IV hydromorphone, considering that Reddy et al is undoubtedly the best/biggest real-patient study looking at this very relevant real-life clinical question. (It also tweaked the IV to PO morphine ratio in a small manner, going from 1:3 to 1:2.5. Probably small enough not to matter for patient care, but enough to make for more challenging math, but that is a small quibble).

I want to observe that I think the Reddy et al study is a landmark study, and I think it provides the best evidence we have of ‘real-world’ “equianalgesia” when going from IV HM to those other opioids. I absolutely think we should all be modeling our clinical practice of opioid conversions on these data as best we can!

The problem is, however, when you try to ‘map’ these data onto the rigidity of an EAT, in which every ‘relationship’ between opioid doses/routes is fixed and bidirectional, you encounter all sorts of problems, ALL OF WHICH exist due to fact that you are using an equianalgesic table in the first place. What I mean is that all the problems that are introduced by creating a new EAT from this study, merely exist because we, as a professional community, have decided that EATs are the best format for presenting information to guide opioid conversions. The fixed and bidirectional nature of EATs itself creates the problem, and my argument is that all this will go away if we discarded EATs entirely as teaching tools/clinical decision aides for opioid switching.

Let’s look at the IV HM to PO MS ratio in the EATs as an example. In the Classic EAT there is a 1:20 (look at the table: 1.5:30 is a 1:20 ratio). Meaning you’d convert 1 mg of IV hydromorphone to 20 mg of PO morphine, and you’d convert 20 mg of PO morphine to 1 mg of IV HM (before dose-adjusting due to ‘incomplete cross tolerance, patient factors, safety, etc).

Reddy et al gave us good reason to believe that when converting from IV HM to PO MS, that 1:20 is too aggressive: the morphine dose is too high. So, if you change the EAT from 1:20 to 1:12.5 which is what the DOC2 EAT recommends, you get a conversion that’s more in line with current knowledge.

So far, so good! But, now look at what happens when you use the tables to convert from PO MS to IV HM. 100 mg PO morphine in 24 h converts to 5 mg IV HM in 24 h using the Classic EAT ratio of 20. Using the DOC2 EAT ratio of 12.5 you get 8 mg of IV hydromorphone. We’ve all of a sudden made the conversion in the reverse direction far more aggressive than it previously was with the Classic EAT! Is there any data to support being far more aggressive in this direction? No, not that I know of; the data are only about going from IV HM to other opioids, not to IV HM.

So, in making a conversion more conservative/more evidence based in one direction, we’ve made it substantially more aggressive in the opposite direction, and done so without any evidence backing it up.

The same thing happens with IV to PO HM. Reddy et al suggests IV HM going to PO HM is a 1:2.5 ratio which is reflected in the DOC2 EAT (Classic EAT has a 1:5 ratio). However, that makes a user of that table use twice the amount of IV HM when going from PO HM than the Classic EAT suggests. This doubling-of-how-aggressive-we-should-be with that conversion, as far as I can tell, is not based on any data, and I argue it’s no small thing to put out a EAT which encourages doubling the dose of IV hydromorphone compared to prior practice without safety data, but it’s an unavoidable byproduct of the Reddy data being forced into the format of an EAT.

My point here is that using an EAT forces us to do this. There is no way around this in the format of an EAT: the fixed bidirectionality of EATs causes this. So—why do we even use EATs?

The best data we have about going from IV HM to PO oxycodone is that it’s a 1:8 ratio. However, the new DOC2 EAT uses the Classic EAT 1:10 ratio. Why? I’m assuming it’s an effort to keep the DOC2 EAT coherent. Think about what it would look like if you used a 1:8 ratio. In the DOC2 EAT, 2 mg of IV HM would map to 16 mg of PO oxycodone. But that means that 16 mg PO oxycodone would map to 25 mg of PO morphine! What a bleeding mess!

And that’s the point: EATs exist in some sort of fantasy universe in which we have for some reason assumed, based as far as I can tell, on no data whatsoever, that the RATIO of morphine to hydromorphone, has to exist in a fixed and eternal relationship with the relationship between, say, oxycodone and hydromorphone. Based on EATs, if you tweak the relationship between oxycodone and hydromorphone it impacts the potency relationship between morphine and oxycodone! They all are forced to shift together because the whole idea behind an EAT is that every value in every cell of that table is ‘equianalgesic’ with the other values.

Why? I don’t see any data to think this has to be the case, at all. All of this strikes me as an overly simplistic and rationalistic assumption of how incredibly complex things work, that creates all of these issues I’ve been elaborating above when you try to do the right thing, which is what Dr McPherson did when she updated her EAT based on good data! The problem is that the format of the EAT itself undermines any effort to use better data to improve it.

Dear colleagues, we don’t have to keep doing this to ourselves. There is another way, some of you are probably already using different methods anyway, and I’ll propose one in a bit, but first I want to make one more argument against EATs in general: they force people into doing unnecessarily complex math.

Part 2: Complex Math is Complex

EATs encourage people to do unnecessarily complicated math. I’m sure I’m not the only greying palliative clinician who’s realized the last several years that a large number of the fellows and residents I work with don’t actually know how to use an EAT. They instead use opioid conversion smartphone apps or websites. Years ago they didn’t know how to use EATs either for the most part, what’s changed now is that they can avoidlearning how to use them completely due to the software converters. (By the way, every single app/website I’ve looked at base their math on EATs and so have in them all the fixed bidirectional problems I mentioned above).

A few months into the fellowship year when I’m working with fellows and one is using an app I’ll ask them to walk me through an opioid conversion calculation. It’s not uncommon for me to realize that they just can’t do it. These are physicians, these are people who are highly educated and know their way around algebra, but they still avoid doing EAT math because it’s clunky, it’s easily confusing, and legitimately easy to make mistakes. This is because they are taught that the way you use EATs to calculate the dose of the new opioid is this:



Holy smokes we make them do cross multiplication!

I’ve learned that cross multiplication makes some people really uncomfortable. And frankly I don’t blame them—it’s easymake mistakes— it’sto multiply when you should divide, etc. I myself, long before my current thoughts about wanting to discard the clinical application of EATs began to solidify, had just completely stopped using cross multiplication in doing these conversions over a decade ago. I have always looked at the table, calculated a conversion factor, and just used that.

For example, looking at the Classic EAT, it suggests 1.5 mg IV hydromorphone is ‘equipotent’ to 30 mg PO MS. That’s a conversion factor of 20: 1.5 x 20 = 30, right? So if a patient was on 10 mg of IV hydromorphone, I’d just multiply that by 20 to get 200 mg of PO morphine, and go from there (adjust the dose down for patient safety reasons etc). I would’t bother doing cross multiplication at all.

This math is much easier to my brain, and also easier to ‘gut check double check’ your math, because you ‘know’ a morphine dose is 20 times the HM dose. With cross-multiplication, those relationships are less obvious, and so easy to invert, especially for novice users, who don’t have an intuitive sense for instance if PO hydromorphone is more or less potent than, say, IV morphine!

Making people to cross-multiplication was a huge pedagogical error. Thankfully, it’s totally unnecessary.

Part 3: Conversion Tables as a Better Tool (?)

All this adds up to my modest proposal: that we do away with EATs entirely as we teach others how to convert between opioids, and replace them with Conversion Tables, that aren’t really tables, just a condensed clinical decision aide in prompting clinicians with easy math and individualized, conservative conversion recommendations that are specific to each conversion. For instance, the conversion tip for going from IV hydromorphone to PO morphine exists independently of the conversion tip for any other conversion including even a different ratio for converting from PO morphine to IV hydromorphone.

I want to acknowledge that I did not invent this idea. I’ll be honest, I’m not sure where in the depths of my memory it came from, but in some ways it’s such an obvious idea I hope when I post this people will comment with examples of opioid rotation decision tools which use the conversion factor approach instead of an EAT! I’m sure it’s out there. If anything is novel about what I’m writing today, it’s mostly in the call to broadly give up on the EAT as a clinical decision tools, not the novelty of using conversion factor tables.

From all this, I’ve mocked up a Conversion Table (pics below of an abbreviated pocket card sized Table and part of the full, annotated Table, with all my explanations and anxious thoughts about the factors, and links to more detailed (but not ready for clinical implementation-see below) versions: Dropbox, Google Drive).



This is one section of the longer, annotated Table:



Benefits and Caveats of Conversion Tables:

• Every single conversion can have its own ratio, and that ratio can evolve/change over time (due to new data or clinical wisdom) easily, without messing up any of the other conversions. You can change the IV HM to PO morphine conversion without that having ripple effects on every other conversion!
• Along those lines, we can have different ratios going from, e.g., IV HM to PO morphine than when we go from PO morphine to IV HM. My argument is that a key reason to do it this way is safety, mainly because if we as a professional community tighten up one conversion it doesn’t force us to ‘liberalize’ its reverse.
• It prompts easier math. Every conversion is just one multiplication or division problem.
• This is true in my opinion of all opioid conversions, regardless of which table you are using: we should increasingly become more and more cautious in all opioid conversions the higher the doses that are involved. I personally think all conversion tables and EATs should have a huge warning on them that says something like “DO NOT USE THESE RATIOS FOR OPIOID DOSES HIGHER THAN 150 MG OF ORAL MORPHINE EQUIVALENTS—CONSULT AN EXPERT” or something like that. (Feel free to argue with me about the 150 mg figure—I don’t myself know what the limit should be other than that number is something that feels good in my gut but that’s all.) I wrote about this in detail in 2019. The bottom line is that I think we should treat all opioid conversion ratios as non-linear at high doses, like we do with methadone at all doses, and that every single guide to opioid conversions needs to SCREAM this.
• Along those lines, I can imagine someone saying about all this, “Look, does it really matter which table we use because, let’s say we’re going from PO to IV hydromorphone, does it really matter if the final answer is 2 or 4 mg!” I would say I am highly sympathetic to that perspective. Look, the reality is, at lower doses, for the vast majority of our patients, it honestly doesn’t make that much of a difference! But, my argument is, the higher the doses involved, the more this stuff really starts making a difference. I am NOT going to lose sleep over if my fellow puts a patient on 2 vs 4 mg of IV hydromorphone a day. I am going to lose sleep over 20 vs 40 mg of IV hydromorphone a day though! This is why 1) I think whatever we put out there for the non-specialist clinician to use needs to be conservative because clinicians will use these at high doses if we put them out there, and 2) we need to signal/emphasize/warn on EATs or Conversion Tables that they are not to be applied at those higher doses! Honestly, I think that’s an even more critical safety practice than fine-tuning the actual conversion ratios.
• Please look through the draft/mock ups of the conversion tables. I don’t mean these to be implemented in clinical settings currently!! I have spent more time arguing with myself about the actual numbers (conversion factors) I put down on paper in those tables than I have writing this cruelly long blog post (you can see some of my comments about my decisions in the Annotated Tables—I’ll let you know that at one point these included conversions around IV fentanyl and I just gave up it’s such a mess). Part of the debate I have with myself is, from a pedagogical standpoint, just how conservative should these numbers be? Should we automatically make them so conservative that we can say that for the ‘average patient’ you don’t need to dose-reduce further? Should we still even be incorporating ideas of incomplete cross tolerance, which, I probably don’t need to tell you, is an incredibly complex, and nuanced thing we inflict upon others. I.e., we show people these tables, tell them to use them, then say, "Oh by the way once you’ve done that you need to make a highly complicated, patient-specific decision about manipulating the final dose ever further, based on these TEN DIFFERENT FACTORS!" I do wonder if we can do better than that.
• I’m not sure what the right answer is here. Certainly, I’m not the one to be deciding this, either! My dream would be that something like this is workshopped and researched with clinical and pedagogical leaders then tested in some way for safety, and I think the ‘best answers’ as to how conservative these numbers should be would come out of that dream process. What I’ve mocked up are things I’d feel comfortable handing a fellow personally, but I just want to be clear to the world I am not proposing my draft tables as anything more than an example of what a hypothetical conversion table could look like in the future. That’s it.

I’m very curious what people think about this. Who’s already moved away from EATs to conversion tables or something similar? What do people think of my draft Tables? I am not an investigator, but if someone who is, and knows how to lead an investigation into validating something like this (not that EATs themselves were ever validated), please ping me. I’d love to chat. I smell a Delphi study!

Post-script, because I sure someone is going to bring this up, and rightfully so: I’m not arguing that we entirely discard EATs for every single application out there, just the clinical/pedagogical ones. They arguably have a role in research and public health—e.g., the ideas in EATs give us a way to calculate OMEs/MEDDs (24 h oral morphine equivalent dose). I think the idea of the OME has its utility at least in research contexts, and we need some way to calculate it, and sure an EAT is an ok way to do that, although I’ll also point out that the OME has been weaponized in the last several years by those who insist, for example, that an OME of 50 mg is some sort of evidence-based hard limit on safe opioid dosing, and the insidious carceral/cops-in-medicine ‘tool’ that is the NarxCare Score.

For more Pallimed posts about opioids.
For more Pallimed posts by Dr. Rosielle click here.



Drew Rosielle, MD is a palliative care physician at the University of Minnesota & M Health Fairview in Minneapolis. He founded Pallimed in 2005. You can occasionally find him on Twitter at @drosielle.

Tweet

by Drew Rosielle (@drosielle)

Several years ago I was giving a talk in front of a mostly palliative care audience and asked, "How often do you find yourself, in the care of your patients, saying to yourself something like 'Thank the stars this patient has a healthcare directive (HCD)'?" The overwhelming answer was a bunch of shrugs and people agreeing occasionally (but not routinely) HCDs are helpful.

For something that seems so central to our work, why is it that so many of my colleagues that day were lukewarm about their utility?

It's fair to say that the last few months have been a controversial one in the hospice and palliative care communities when it comes to ACP, triggered by Dr Sean Morrison's editorial in Journal of Palliative Medicine: Advance Directives/Care Planning: Clear, Simple, and Wrong.

I've been mulling over that editorial for a couple months now and then noticed that, arguably, one of the most important advance care planning (ACP) studies in recent history has just been published, although probably not for the reasons the investigators were hoping for, and I thought I'd write about where we're at a bit for Pallimed.

This post is going to be about my own thoughts as to why I mostly agree with Dr Morrison's editorial. The next post will be about this tantalizingly important paper that was just published.

If you haven't read Dr Morrison's editorial, please read it now--it's not that long and it seems to be open access at the moment. Here's his top-level argument:

As of May 2018, 80 systematic reviews of advance care planning covering >1660 research studies had been published and the National Institutes of Health and the Patient-Centered Outcomes Research Institute had funded >750 research studies at a cost of >$300 million in taxpayer money (an average of $1 million/year). Foundations have spent untold millions. What evidence has all of this research and money wrought? The >1660 studies and 80 systematic reviews on advance directives/care planning have produced only limited and low-quality evidence that advance directives/care planning can actually result in changes in patient, family, clinical, utilization, or financial outcomes. Like the study published in this month's journal, the majority of high-quality research has failed to find a strong link between advance directives/care planning and outcomes of importance. Similarly, despite 30 years of diverse interventions and programs, two-thirds of adults do not have an advance directive (an imperfect yet reasonable surrogate for occurrence of an advance care planning discussion).

I mostly agree with Dr Morrison about this, and will flesh out my thoughts as to why. I need to note that I didn't run any of this by Dr Morrison so any claims I'm making here about what he thinks are my interpretation of his editorial, sort of my own internal conversation with his editorial, and nothing more; I'm not claiming to speak for him. Also--JPM published many letters in response to his editorial, all of which are worth your time reading.

I.
Takes like Dr Morrison's are really controversial (in part) because I think people (clinicians in HAPC), use different definitions of ACP. I sometimes think when people hear stuff like "ACP is not helpful" they think people like Dr Morrison is saying something like "Discussing prognosis and planning for future care with your patients with advanced and serious illnesses is worthless." I.e., this is sort of a 'Big Tent' definition of ACP which includes activities associated with completion of health care directives + really any clinical activity that's in the realm of discussing the future (eg discussions of prognosis, preparing patients for future decisions/what is likely to happen to them in the future, even goals of care discussions). I'm pretty sure though that Dr Morrison and the other detractors of ACP, when they say there's no evidence it actually helps patients, are really talking about the narrow definition of ACP (completion of HCDs and the clinical/healthcare discussions that occur adjacent to HCD patient education and completion), and are not criticizing the idea we should be discussing prognosis / the future with our seriously ill patients.

Also, because I know some of you will be reading this and asking, "But what about Physician Prders for Life-Sustaining Treatments (POLSTs)?"-- for all intents and purposes the following discussion is not really about the POLST movement (which is not to say POLSTs have the sort of evidence behind them we'd like either, but that's another story). All this isn't really about Ariadne's Serious Illness Care project either, and importantly I'll note the Ariadne investigators are doing their best to study their intervention in multiple settings and are not advocating wholesale population wide implementation of their program in the absence of evidence (which is what was done with ACP the last decades).

II.
One could note that there really isn't great evidence for talking with seriously ill patients about prognosis/planning for future care either. That is sort of true, there is some research but not enough and not overwhelmingly convincing (one could look at the recent Ariadne Serious Illness work for some examples of what I mean by sort of true) but honestly I'm less 'worried' about that because as a physician I have a professional obligation to discuss prognosis with my patients. I can't help them make informed care decisions otherwise, and honestly I'm less interested in outcomes research focused on if I should be talking with my patients about the future than how I should be talking with them about the future. I want to learn better/more effective/more compassionate/more accessible/more timely/more culturally- and patient-centered ways of having those conversations, not whether or not I should be having them.

At the same talk I mentioned above, for instance, when people admitted they rarely have found ACPs to be helpful in real clinical practice, I asked people how often are they glad that someone before them had had a realistic discussion with a patient about the future, and everyone indicated that they found this routinely helpful of course.

III.
The same is not true with narrow definition ACP (and from here on out when I say ACP I am only talking about the narrow definition). ACP was adopted as a healthcare innovation really without the sort of scrutiny and demand for proof that it actually helped patients that we'd demand for other innovations like, say, LVADs. The acceptance of ACP as unimpeachably a good, obvious, thing for our patients is so entrenched that people have funded entire careers off of studying how to get people to do ACP, without there being reasonably clear evidence it actually improves patient-centered outcomes.

Much of the ACP research out there is focused on projects/methods to get patients/clinicians to do ACP. I'm not claiming this is bunk research, some of it is really high quality work, however I am claiming that it's high quality work which seems to assume we are benefitting patients by helping them do ACP yet we don't actually yet know that. Importantly here, if you are reading this and thinking to yourself something like I just saw a patient the other day who seemed to benefit in some way from ACP I am not saying you're wrong. I've met that patient too. But, like with anything, acknowledging there are a few patients which benefit from something is very different than concluding something should be a routine, integral part of medicine, implemented widely as a public health intervention, and worth the immense amount of healthcare and research investment of time and dollars spent on it.

I think Dr Morrison's thesis is "ACP is not worth the time/money/effort that has been spent on it as a public health intervention," not "ACP has never helped anyone."

IV.
Some ACP research has been about 'congruence,' whether doing ACP activities predicts whether you will later on get the care/treatments you at a prior date said you would want. Especially in the early days of ACP most of the talk was about 'autonomy' and making sure people get what they want, etc. (Indeed, much of the early talk about ACPs was really centered on things like patient rights and autonomy.) There is a moral and philosophical discussion to be had about that topic and whether this is something we should actually be trying to achieve (because patients wishes change over time, and how can patients really make informed decisions about future health states without actually being in such a state, and does it even make sense to ask patients to sort of 'pre-decide' healthcare decisions which on a good day are bleedingly complicated even for cognitively intact and emotionally stable patients). It's an interesting topic, but it doesn't really matter anyway, because there's no high quality research into congruence showing ACP does anything anyway.

To take a couple examples, there is this very well done Australian study from 2010 in the BMJ (the "Melbourne Study") which did a comprehensive ACP intervention in high risk hospitalized older patients. If you read the abstract you'll think that the intervention improved 'congruence' (meaning patients who did the ACP intervention were more likely to get congruent care than those who didn't). If you actually read the study you'll realize that what the intervention group had was better documentation of their wishes so it was easier to 'prove congruence for them (we don't know if the control group had 'congruent' care because their wishes weren't documented) and we don't actually know if the ACP intervention changed any patient care. The 2nd example is the only RCT that I'm aware of the highly revered Respecting Choices ACP model which looks at congruence. This was a RCT of 300 high risk patients (severe heart failure or were on dialysis) who received usual care or the Respecting Choices intervention. This carefully done prospective trial which followed patients long term showed absolutely no improvement in congruence. Think about this one--this wasn't a study of healthy elders, this was a study of people at really high risk of dying (ie the people most likely to benefit from ACP), 300 of them, and yet they couldn't show any difference. It's research like this that Dr Morrison is pointing to I believe when he says ACP has not been shown to benefit patients.

V.
Many of you are probably aware though of all sorts of positive ACP research. People who have HCDs are, compared with those patients who don't have them: more likely to have DNR orders at the end of life, more likely to spend more time enrolled in hospice, less time in ICUs at the end of life, etc. We have a couple decades of research showing this in multiple settings. The problem with this is that nearly all this research which has shown these lovely associations is observational, subject to major confounding. And there is every reason to believe there is huge confounding at play in this sort of research. That is, patients who are open/inclined towards participation in ACP activities are probably more inclined to want 'less intensive' EOL treatments. Think about your own patients, there is a group of them for whom participating in ACP is pretty easy, they want to do it, it's emotional for them but they are committed to it without much encouragement from us, etc. I think there's every reason to believe that the average patient like that is also far more disinclined to want 'invasive' EOL treatments than patients who will not participate in ACP. What this ends up meaning is that having a HCD may just be a proxy for also being disinclined to want 'invasive' EOL treatments, and the actual process of completion of the HCD doesn't impact care one way or the other because these patients already are the folks who are inclined to enroll in hospice earlier, etc. Again, when you do prospective, controlled research, it's much more difficult to show any patient-centered 'benefit' or change in outcomes, apart from the fact that more ACP is documented, even when you are using one of the most highly investigated and validated ACP tools out there, such as in this trial.

VI.
I disagree with Dr Morrison on one major point--he seems to be advocating that we wholesale abandon ACP as a clinical and research endeavor. I think there is one domain of patient outcomes which is really under researched, and ACP may really help. This is the possibility that ACP's main benefit to patients/families is essentially a palliative benefit to reduce iatrogenic emotional suffering we in healthcare cause our patients/families due to our decades' long abandonment of our dying patients by forcing them to make decisions they don't want to make. Because we have moved in the last few decades to a situation in which we have decided that desperate, grieving family members are the defacto best people to decide, for instance, if and where tubes should be inserted into their dying family members' bodies, we have caused tremendous suffering, and ACP may be a way of mitigating that suffering. I think this idea has potential and it has increasingly been a subject of investigation in the last decade, which I'm glad to see, although it's still not being investigated routinely. For example the Melbourne Study, while it didn't clearly show any other benefit, did show decreased stress and anxiety symptoms in family members who participated in the ACP process.

This is presumably what Ira Byock meant when he said "I have an advance care directive not because I have a serious illness but because I have a family.” Ira, of course, might be on to something there, and it's been frustrating to me that so few investigators have taken up that outcome as a primary focus of ACP. Admittedly, it's a bit obscene to say it's just iatrogenic suffering here which ACP might mitigate, of course it isn't, and there may be emotional and social benefits outside that--families talking about what's important together, memory and legacy making that can occur adjacent to ACP completion, etc. What I am essentially proposing is that ACP could be studied in a way that's sort of similar to how the Dignity Therapy investigators approach their intervention. ACP is not Dignity Therapy, however I'm arguing it could be investigated as a sort of analogous suffering-reducing, meaning-making intervention.

Importantly, I am not proposing here that we just shift uncritically from saying ACP is good not because it's all about 'honoring'/'respecting' choices (the pre-decisions, autonomy model), or because it reduces unnecessary healthcare costs / keeps people out of ICUs when they are dying, but because it reduces suffering. It's premature to say that. We can't put the cart before the horse. ACP may yet be worthless as a broadly enacted healthcare intervention, and we should continue to consider any benefits hypothetical until proven otherwise

For more Pallimed posts about advance care planning.
For more Pallimed posts by Dr. Rosielle click here.

Drew Rosielle, MD is a palliative care physician at the University of Minnesota & M Health Fairview in Minneapolis. He founded Pallimed in 2005. You can occasionally find him on Twitter at @drosielle.

Tweet