Examining the Pharmacoeconomics of HIV Treatment
Examining the Pharmacoeconomics of U.S. AIDS Drug Access, Washington, DC, April 6, 2004
I am quite honored to come and deliver a quick presentation for you on the pharmacoeconomics of HIV treatment. First, a little bit about my background: In addition to teaching at the University of Missouri-Kansas City, I practice at the Kansas City Free Health Clinic, which is a completely free-of-charge clinic with lab and medical services. Everything is provided free-of-charge to anyone who walks in the door. We require neither proof of insurance nor of any financial source.
The guidelines for the administration of antiretroviral drugs in adults and adolescents were recently updated March 24, 2004. Many of you are familiar with the US Department of Health and Human Services' "Guidelines for the Use of Antiretroviral Agents in HIV-Infected Adults and Adolescents" (DHHS Guidelines), but I took no chances in the event that some of you were not familiar. Because I do teach most of the time, I have two reasons to use these guidelines: 1) to define our goals of therapy; and 2) to illustrate the tools we use to achieve the goals of therapy.
I am going to address several issues through economic analysis. First, how to obtain maximal and durable suppression of viral load, below the limit of detection, whatever that definition may be at the time; also, how to restore and preserve the immune function. We may not be able to fully restore or bring patients back to where they once were, but we can prevent further decline of their immune function. I will discuss improving or at least not completely losing quality of life. Finally, I will analyze the incredible reduction in morbidity and mortality that has been shown with these antiretroviral agents.
In order to get there, we need to maximize adherence to an antiretroviral regimen. I will discuss the different components and the thought processes that go into constructing one of these regimens. I am going to try to give you some insight into that based upon something that happened about 24 hours ago. Preserving future treatment options may be an answer to the question that was raised earlier: "Is an economic perspective going to be used in developing treatment guidelines in the future?" I think that as I go through my presentation you are going to find the answer to that question is absolutely "yes, it will be." One thing I am also going to touch on at the very end of my presentation is the use of drug resistance testing in clinical practice.
Slide 1 shows the four classes of currently available antiretroviral agents. This is fantastic. Ten years ago that was not the case; two years ago that was not the case. We now have the non-nucleoside reverse transcriptase inhibitors (NNRTIs), the protease inhibitors (PIs), the nucleoside reverse transcriptase inhibitors (NRTIs), and just coming on board are the fusion inhibitors. Below this, as you will see, are two excerpts from Table 12A of the DHHS Guidelines, which recommend the "preferred regimens" to start in persons who are HIV infected and meet criteria to start therapy. If you pay close attention to some of the previous presenter's slides, she talked about pricing based upon preferred and non-preferred agents. If you look closely at the DHHS Guidelines, you find exactly the same wording. There is a preferred regimen and an "alternative regimen." If you are a bean counter by trade, you can very quickly take that wording and apply it to how much you are going to charge your patients if they are on a preferred regimen versus an alternative regimen.
I want to point out that regimens are broken down by class of antiretroviral agent. There are NNRTI- and PI-based regimens, each regimen consisting of at least three agents. What does this mean for pill burden for the purposes of costing? There are now co-formulations available as one drug -- such as Kaletra®, which is the co-formulation of lopinavir (LPV) and ritonavir (RTV). Taking into account a preferred PI-based regimen -- LPV/RTV + lamivudine (3TC) + stavudine (d4T) -- we are talking about three total drugs in an actual combination. If we look at an alternative PI-based regimen, for example, indinavir (IDV)/RTV + 3TC + d4T, we are talking about four total drugs in an actual combination -- because none of these antiretroviral agents is co-formulated. A previous presenter mentioned that caps are being imposed on the number of prescription medicines that can be filled per month. We need to have at least three drugs on board to construct an effective regimen. So, if you are limiting prescriptions, and you are counting co-formulated drugs as one prescription, the co-formulated drug would automatically move higher up on the preferred list.
I saw a fellow by the name of Bill yesterday. He was diagnosed in October 2003. His CD4 count was 368 cells/mm3. His initial viral load was 113,000 copies/ml. We did not start Bill on therapy initially, and the student who was with me yesterday was very puzzled... She said, "You have a person here who obviously needs to be on therapy. Why haven't you started him on therapy before now?" What I asked her was to go back and look at the other medications that he is on. There was this thing about Bill's acute schizophrenia that we had to get resolved before we could start him on medicine. While it is fantastic to put patients on antiretrovirals and get their viral load undetectable and their CD4 count up, if they are schizophrenic or psychotic, we do not really want to have them out there. So we prioritize what we are going to treat first. Bill, however, was at great risk for the development of AIDS. His viral load was 113,000 copies/ml at baseline. When he showed up yesterday, by the way, his viral load was 450,000 copies/ml and his CD4 count had dropped to 178 cells/mm3, but he was stable on his anti-psychotic meds so we were very happy about that.
Slide 2 shows the proportion of individuals who have a CD4 count of less than 200 cells/mm3 and viral loads in different ranges who are going to go on to develop an AIDS-defining illness or die due to an AIDS-related illness over the course of three, six, and nine years. We use this table to decide who should start therapy at any given time. In this particular instance, without question, it would be time to start therapy, but although our patient met these criteria, we decided not to begin therapy because patients are individual human beings. I have a number of colleagues in pharmaceutical sciences and pharmacology, and I give them a difficult time whenever they tell me they do not understand why, when I show up in the lab, my experiment is not ready to go. I tell them, "Well I do not deal with rats and I do not keep them in cages in the lab, so if I show up it does not necessarily mean they are going to show up that day." It is one of the things I try and emphasize to my students, as well as audiences, that these are humans with whom we are dealing.
There are guidelines and there are variances within the guidelines. In Slide 3, we can see where Bill falls in terms of risk for developing an AIDS-defining illness in three, six, or nine years. Slide 4 is a graphical representation of these data. This figure is modified from the graph in the back of the DHHS Guidelines, and it shows the four strata based on HIV RNA viral load: greater than 55,000; 20,000 to 55,000; 7,000 to 20,000; and 1,500 to 7,000 copies/ml. CD4 count range is also shown. As you can tell, as the CD4 count continues to decline, the likelihood for developing AIDS within three years significantly increases, and this has a major influence on what we decide to do and how aggressive we are in doing so. A side note, when we saw Bill yesterday, he was there for two and a half hours to get started on his antiretrovirals. Not only did he see me, he also saw the physician, and then went on to see his case manager, his substance abuse case manager, and his peer counselor. If Bill had a job he probably got fired because of how long he was at his doctor's appointment yesterday.
When we start someone on therapy, one of the things that health maintenance organizations (HMOs) and other healthcare payers look at is lifetime cost to treat these individuals. Along the X-axis in Slide 5 are CD4 counts. If we look at this as a one-dimensional approach only, the greater a person's CD4 count at initiation of therapy, the higher the overall lifetime cost to treat, using just antiretrovirals, without any other factors. Costs of starting therapy at CD4 counts greater than 500 cells/mm3 are significantly higher compared to waiting until the count drops to 400, 300, and 200-299 cells/mm3. Costs for starting therapy at less than 200 cells/mm3 are elevated mostly because patients are more likely to be hospitalized. If we take this figure alone and you want to decide to develop guidelines of when you are going to start someone on therapy, then it looks like 200 to 299 cells/mm3 is the optimal time to start therapy. However, the perspective I prefer is that once we start therapy, what is the cost to treat per month lived?
In this instance what we find is that the graph gets reversed (Slide 6). Those patients who were started on therapy with CD4 counts of greater than 500 cells/mm3 over the course of their treatment lifetime were cheaper to treat per month. If we take a cost per month approach to treating these individuals, we can see that it is better to treat them earlier on, expanding more access to antiretroviral agents. So if they present initially with CD4 counts under 200 cells/mm3, since we see the cost to treat per month lived is considerably higher than any of the others, should we just go ahead and not offer them therapy? Most of us would absolutely not agree to this. Even if their CD4 count was less than 50 cells/mm3, if they already had an AIDS-defining illness, we know that individuals on antiretroviral therapy cost the system substantially less than those individuals who had not been started on therapy.
Similar data were broken down and published in the New England Journal of Medicine about three years ago (Slide 7). They looked at the initial CD4 count of the presenting individuals, the lifetime cost to treat these individuals, the life expectancy, and the incremental cost per life-year gained, broken down by CD4 counts of 50, 200, and 500 cells/mm3. They also looked at those individuals who were on no antiretroviral therapy versus a three-drug regimen, not specifying the regimens. Total lifetime costs were higher for those individuals with higher CD4 counts, again, reinforcing the data that I previously showed you, and as CD4 counts decrease, the lifetime costs also decrease. Overall, lifetime costs are greater for people who are on medications if you look at the cost alone.
But when we adjust for quality of life, we can see the incremental cost per year of life gained is less as we treat these individuals earlier on. As we get them stabilized, we decrease viral load, we improve CD4 counts, and we improve immune function. The earlier we are able to do this, the more of a benefit we are able to give our patients. That was a generic three-drug regimen approach, but much of the data that I use are based on PI-based combinations, because those agents have been out for a longer period of time, more researchers have conducted these types of cost comparisons with these particular agents and there are more data currently available.
What we can see in Slide 8 is that if we are able to achieve an undetectable viral load for three years, the cost benefit of PI-based therapy is over US$5,000 per life-year gained. And if we can delay AIDS for 20 years in these individuals, as many of us are trying to do, and make HIV a chronic disease, that is a worthwhile goal to have. We can also show substantial cost savings to an institution. Slide 9 reflects data from four recent trials -- ACTG 320, the Johns Hopkins HIV Clinic cohort study, the INCAS study, and the Dupont 006 trial -- in which we see that those people taking potent antiretroviral drugs are costing the system more if you take that unique directional approach of just looking at one aspect of this. But we can see life expectancy is greater, even when adjusted for quality of life, and the cost overall was much lower in the individuals who received triple-drug therapy. One of the things noted is that overall, PI-based regimens were more expensive than NNRTI-based regimens. That is a generalization, but it is also something that is actually still holding true at this time.
The cost benefit for PI-containing regimens, if we look not just at outpatient but also inpatient expenses, and if we look at the average length of stay for individuals who are receiving a PI-based regimen, we can see a decrease in the number of days in hospital for people on therapy versus those individuals who are not on therapy (Slide 10). Hospital costs decrease substantially for those individuals who are on therapy versus those that are not. This is to reinforce the data I previously presented about how these drugs, when started, can not only impact a person's life, but can also impact the savings to the institution no matter what stage of the disease treatment is started.
If you look at the change in per patient per month costs as we increase use of PIs, we will see outpatient oral medication costs go up, but that is more than offset by a decrease in hospital costs, professional costs, lab costs, and home healthcare costs (Slide 11). For every 10 percent increase in PI use, there is an average increase of about US$135 when starting individuals on these therapies.
I have been discussing PIs and NNRTIs, and I have tried to stress that it really does not make a cost difference -- Slide 12 illustrates my case. A PI-based regimen -- IDV + AZT + 3TC -- was compared with an NNRTI-based regimen -- efavirenz (EFV) + AZT + 3TC. What researchers found was that the average cost between the two regimens was not significantly different for either experienced or naive patients. Despite differences in viral load, CD4 count, and adverse events, whether a person was diagnosed with HIV or AIDS, researchers really were not able to tease out a difference between these two regimens. Both regimens were shown to be very cost effective.
It is not just getting people started on therapy that we must strive for to reduce costs to our system, it is getting them and keeping them engaged in primary care. As I mentioned earlier, Bill was meeting with a number of individuals for two and a half hours yesterday. The reason is that if we are able to start and keep patients engaged in care in our clinic, we are much more likely to achieve success. Some in-house analyses that we have done have shown that patients are more likely to get to an undetectable status if they are able to come back for at least two follow-up appointments after starting their antiretroviral therapy. That does not sound like a lofty goal but that is where we start and we are trying to improve from there.
In Slide 13 when we look at the length of stay, the cost in millions of US dollars, the number of emergency room visits, and the number of hospitalizations seen on an annual basis, individuals remaining in clinical care have much lower numbers than those individuals who started on therapy but dropped out of clinical care. Engagement and continuation of engagement in primary care is paramount to the success of antiretroviral therapy. These drugs can be as potent as anything else in the world, but if the patients do not take them you can rest assured they will not work.
One of the other expenses for primary care, however, and it is not an insignificant one, is the cost for laboratory procedures. Slide 14 shows just a very small sampling of the types of labs that we do and the percentage of abnormal results that we see in our patient population. Each of the drugs has criteria that need to be monitored. Liver enzymes, renal function, anemia, hematocrit, and white blood cell count must all be watched closely and followed up as necessary. This does not include resistance testing; it does not include a number of other tests that are now being used but is simply to show you that we do check a number of laboratory parameters in our patients. We do a number of procedures in them and many times we find that the results are abnormal. In our clinic population, abnormal results are quite different than what you might find in others.
Though we do all of the labs every three months, or even more frequently if, for instance, they were not fasting when they were supposed to be fasting, the lab is not a significant component of the overall cost when providing primary care for an individual. This is true whether the patient has failed an antiretroviral regimen more than one time, has only failed one regimen, or has never failed a regimen. I say that tongue-in-cheek because it is not the patient who fails the drugs, it is the regimen that does not work.
What we see is that as individuals fail therapy, the cost for providing primary care continues to increase (Slide 15). So do we decide that, since they are failing therapy, and these are the patients who are costing us the most money, we will not provide antiretrovirals to these individuals? We do not make that decision, because, again, it is more than a one-step approach. It is more than a unilateral approach to looking at the cost. If we look at the mortality of individuals who have low CD4 counts but who could not achieve undetectable viral loads, there is a substantial difference between those who have been receiving antiretroviral therapy and those who are not receiving antiretrovirals. If you look at the incidence of opportunistic infections in these different groups, keeping patients on a "failing regimen" was proven to provide a substantial improvement in those individuals from a cost standpoint. The cost of one admission, actually the cost for an emergency room evaluation to treat pneumocystic pneumonia, will pay for an antiretroviral regimen for an entire year for one individual. It has been shown numerous times that preventing these opportunistic infections is paramount in overall cost savings.
I did say I would talk briefly about resistance testing. Slide 16 offers an analysis by Kit Simpson [Medical University of South Carolina] of the cost of resistance testing using three different models to determine whether resistance tests would ultimately prove cost-effective. The reason I cite this work, even though this group has gone on to show different results based upon different trials, is that depending on which methodology was used, they came up with either a benefit, a cost, or break-even. Unlike many of the colleagues with whom I converse on a routine basis, I am not a great believer in resistance testing, and it probably has to do with the limited amount of resources we have at our disposal. I guess I am just jealous of those sites that actually are able to do resistance tests, since I just have to do without them.
There was a question earlier about the actual cost for antiretrovirals, and I do not necessarily believe in using the red book, which most people use as the standard in determining the annual average wholesale price of a medication in the United States. This is because I found it to be much more competitive if you do like the rest of the world and you get on E-bay and you ask: "How much does this cost?" So these are E-bay costs. This is how I can get my medications through E-bay. I do not know if it is legal or not. I have not actually tried to buy them ... because I am sure I would be in trouble in three different states.
Slide 17 shows the annual cost for the PIs. For those of you who are engaged in doing this on a routine basis, if we had those 60 individuals who were able to take ritonavir (RTV) in the original RTV trial, and they were on 600 mg of RTV twice a day, the cost for their PI alone would be about US$46,000 a year. Now most of us have used RTV to boost, or pharmacokinetically enhance, other agents to improve the activity of the other agents. We can see that the cost is still substantially higher, and this is not even using the drug at its full dose. This is 200 mg a day. Again, as was mentioned earlier, in some states there is a cap on the number of drugs, so you can see why a state such as Texas would want to cap RTV at 200 mg a day because the cost of RTV 200 mg a day is over US$30,000 a year.
Slide 18 shows the cost for one NRTI. For example, the price of emtricitabine (FTC), a recently approved once-a-day drug, is running just over US$3,000. Combivir® is prorated as though it were two agents, so Combivir® is actually twice this price because it is actually two drugs in one. So we can see that the average price for one NRTI is a little over US$4,000 per year.
Slide 19 shows two NNRTIs -- EFV and nevirapine (NVP). Efavirenz costs a little over US$5,051 a year and NVP costs about US$500 less per month; you multiply that times 40 million individuals who may be on antiretroviral therapy and that is a substantial cost savings.
Slide 20 shows the annual cost for an antiretroviral regimen, this is just the antiretroviral. It is not the labs. It is not providing case management, substance abuse case management, or any of the other ancillary services that I talked about. It is not hospitalizations. This is roughly about US$20,000 to US$25,000, and takes into consideration the average price of a PI, the two NRTIs, and an NNRTI. The reason I added all of those up is because that is actually the regimen that we started Bill on yesterday -- all four of those agents from all three classes of available antiretrovirals. We have the temporary benefit in Missouri of covering all approved antiretrovirals. I believe the number of drugs available through Missouri's AIDS Drug Assistance Program (ADAP) is second only to New York State. Because of that, we were able to stabilize Bill on his anti-psychotic medicines through ADAP, and we now have him on antiretroviral therapy.
What I have tried to present today is some brief background -- some of which is published, much of which has been presented elsewhere. Finally, I think Abbott Laboratories has brought to the forefront with their RTV pricing increase last year exactly what is going to be coming along in a few short months to years with antiretrovirals.
Patrick G. Clay is Assistant Professor of Medicine in the Division of Pharmacy Practice at the University of Missouri in Kansas City.
Back to the May 2004 Supplement issue of IAPAC Monthly.