February 9, 2009
My name is Steve Berry, and I'm with Johns Hopkins University. I'm presenting the results of a prospective observational study showing the reasons for hospitalizations in HIV-infected patients after HAART initiation.
We looked at 2,000 people in our center in Baltimore, Maryland who were previously naïve to antiretrovirals. All of them initiated HAART between 1996 and 2005. We looked at their hospitalization rates within six months prior to starting HAART, and then looked at all hospitalizations over time in that first year after starting HAART.
Stephen Berry, M.D.
The major findings: number one, was that all-cause hospitalizations remain high for the first 45 days.
However, by 45 to 90 days, and then out through one year after starting HAART, they are statistically significantly less than prior to starting HAART.
Then we looked just at the baseline characteristics of our group. The group is about 66 percent men. The median age for starting HAART was 39. The group was about 75 percent African-American. About 45 percent have injection drug use as a risk factor for HIV infection. The bulk of the people's HAART initiation events that we studied were actually occurring prior to 2000, with a large number, 54 percent, being 96 to 98 -- shortly after HAART came out.
When we look at all the categories for reasons for hospitalization, the top category was actually not AIDS-defining illnesses, but were non-AIDS-defining infections. This included episodes of pneumonia; this was bacterial endocarditis; this was cellulitis -- as the top types of infections in that category.
The next category for hospitalizations was AIDS-defining illnesses, where PCP [pneumocystis pneumonia] was top, accounting for 25 percent of those; Cryptococcus was next; candidal esophagitis was next.
Psychiatric illness was actually the third most common reason for hospitalization in this overall period, accounting for about 11 percent of all admissions. Gastrointestinal and liver causes were next. Endocrine nutritional, metabolic and immunity, as one combined category, was next. The bulk of that was hypovolemia as a reason for admission; then reno- and genito-urinary; and then cardiovascular, to round out the top seven categories for most common reasons for admission.
What we did next was to look, for each of those categories, over time after starting HAART: what happens to admission rate within each of those categories?
So, for the two reasons for admission which were most common -- the non-AIDS-defining infections and AIDS-defining illnesses -- we see the pattern -- which is what we see for all-cause admissions, and what was presented in October -- that for the first 45 days, admission rate remains quite high.
Then it seems that, at 45 to 90 days, the real immune recovery has kicked in and is operating, and you see decreases for all these manner of infectious illnesses.
Next, looking at psychiatric reasons for admission, there's also a statistically significant drop-off. This drop-off occurs immediately after starting HAART, and remains constant out through one year.
When we look at all of those other categories for admission -- GI [gastrointestinal], liver, endocrine metabolic, renal and cardiovascular -- there doesn't appear to be any change after starting HAART. Admission rates within those categories remain flat across time.
We did multivariate analysis of admission rates for the top three categories -- so, for non-AIDS-defining infections, AIDS-defining illness and psychiatric illness. And here, as you would probably expect, for AIDS-defining illnesses -- CD4 count, baseline CD4 count, is the largest driver. But the decay over time is not at all affected in multivariate analysis. Still, after 45 days, you see this decrease in admission rate.
And the greatest risk was ...? I see you had a lot of patients with below 50-cell CD4 count at baseline.
Definitely. In that risk group, there was an incidence rate ratio of infection of over 10, if you had less than 50 cells, compared to having over 200 cells when you started HAART.
And did you track the rise in their CD4 count, and correlate it with hospitalizations?
We didn't have a chance to do that. We didn't track that.
Are there plans to do that?
We may look at that. In fact, we may look at that especially, as well, because there's a lot of interest now in looking at some of these non-AIDS defining illnesses over time, after starting HAART, and even after starting and restarting HAART, looking at cardiovascular events, for example, and risk of MI [myocardial infarction], in particular. So, using CD4 rebound in that analysis would be something interesting to do.
Could you talk about the psychiatric illnesses? Because it's not something one thinks about when predicting reasons for hospitalizations for HIV-infected patients.
Sure. I think the first major finding is that they represent a significant burden of illness. Overall, they represent 11 percent of reasons for admission. They do have that drop-off in admission rate right after starting HAART. What's interesting here is that we can't tell from these data whether the decrease in psychiatric admissions actually causes people to start HAART, or whether HAART, and getting better from HIV, leads to a decrease in mental illness. I actually would suspect it's the former. My suspicion is that people who are getting admitted for depression within the six months prior to starting HAART: a lot of them are getting engaged in care, and they are getting better. As part of that getting engaged in care, their providers are recognizing these patients have low CD4 counts and that HAART is indicated. So then they're starting HAART. So it's hard to tell in this case which came first.
One little piece of corroborating evidence is that the drop-off in admission rate for psychiatric illness happens immediately in the first 45 days after starting HAART. I think that makes me suspect that it's actually improvement in mental illness is preceding the HAART initiation.
And the most likely patient to experience psychiatric hospitalizations?
The most likely patient, we can see here. It's going to be someone who's been using injection drugs. Women were also more likely in multivariate analysis to be admitted for psychiatric reasons. Interestingly, younger age, rather than older age, was also significant in multivariate analysis for being admitted for psychiatric reasons.
Have these numbers been corroborated in other studies that you know of? In terms of the psych? Psychiatric illnesses are very interesting. I haven't really heard that before.
In terms of psychiatric illness itself, I don't know the literature. But we know that for overall reasons for admission, injection drug users [IDUs], women, African Americans, are all more likely to have admission, and are also more likely to die.
But the difference is very striking here. I mean, for psychiatric illness admissions, it's really almost double, even between women and IDU.
Well, we do see for non-AIDS-defining infections, there's also, similarly, a higher risk for women and a higher risk for IDU. It's logical for IDU, because substance abuse correlates very strongly with depression. In these admissions, overall admissions for psychiatric illness: recurrent major depression was the number one cause; depressive disorder not otherwise classified was the second one; and drug-induced depression was the third one. So I think a lot of what we're seeing here is comorbid illness with substance abuse and depression.
Do we know what sort of drugs the IDU were using?
Primarily heroin. Cocaine is often combined, injected at the same time.
But actually, this is a good story, as well as a bad story. The good story is that there's a lot of difference within 45 days.
Exactly. So 45 days is a period that clinicians may want to keep in mind. It represents a period of the time when you really need to be quite concerned. But then after 45 days, making it through that high-risk period, we see great improvements across the board.
Great. What are your next steps?
The next steps for this would be, as mentioned earlier, looking at people who are not naïve to antiretrovirals, looking at people who are starting and stopping HAART, and looking for causes of admission in those groups, maybe paying particular attention to some of the non-AIDS-defining illnesses. Looking at psychiatric illness again, for example, looking at cardiovascular illness, looking at renal illness.
Great. Well, thank you very much.