Why Do People Taking Antiretroviral Medications Seem Sicker Than People Not Taking ARVs? Do ARVs Make You Sicker?
People often write in saying "HIV positive people who are taking ARVs seem to be sicker than HIV positive people who are not taking ARVs. Therefore, I think that HIV is not the cause of AIDS, and/or that ARVs make you sicker and are actually causing the symptoms that doctors say are symptoms of AIDS." We once received an email from a person who had counted their HIV positive friends who were on ARVs and not on ARVs and compared their health in detail over time, concluding that their "scientific study" of the two groups had shown that people on ARVs are sicker than people who are not taking ARVs, and that therefore ARVs are harmful.People who draw this conclusion are making a basic but understandable mistake: they are comparing apples and oranges. In standard medical practice today, ARVs are usually prescribed in the later stages of HIV infection (typically, to people whose CD4 count has fallen below 350 and/or have developed an opportunistic infection indicating that their immune system has grown weaker). Therefore, HIV positive people who need to be prescribed ARVs are by definition sicker (further along in the course of their HIV infection, needing help from ARVs to stay healthy) than HIV positive people who do not need to be prescribed ARVs.
Therefore, if you want to determine whether ARVs are helpful or harmful for people who are HIV positive, you can't compare the health of HIV positive people who are on ARVs with the health of HIV positive people who aren't on ARVs because that would be comparing apples and oranges. "People not taking ARVs" are not a matched control group for "people taking ARVs" because the two groups are have a basic difference: "people taking ARVs" are further along in the course of their HIV infection than "people not taking ARVs."
It may be easier to see the mistake people are making if you think about two other conditions people are more familiar with: coronary artery disease and diabetes.
Example: Coronary Artery Disease
Having high cholesterol is a risk factor for developing coronary artery
disease and having heart attacks. In the later stages of coronary
artery disease, fatty plaques build up within the arteries that supply
blood to the heart. This increases the risk of a clot blocking one of
those arteries, causing a heart attack and possibly leading to death.People who have advanced, later-stage coronoary artery disease may be given a heart bypass operation in which veins from the leg are surgically grafted on to the heart to act as "backup" arteries that carry blood around the arteries that are nearly blocked. Having a bypass operation may reduce your risk of a heart attack and death if you have advanced coronary artery disease.
If you were trying to understand whether bypass surgery is helpful or harmful, you might look at all your friends who have high cholesterol and compare "people with high cholesterol who had bypass surgery" with "people with high cholesterol who didn't have bypass surgery." Those who underwent bypass surgery would have a fair number of heart attacks and deaths. Those who don't have bypass surgery would have fewer heart attacks and deaths. Does it follow therefore that bypass surgery causes heart attacks and deaths? This is precisely the (incorrect) conclusion you'd reach if you took this approach.
But no, that's not what it means. It means that if people have high cholesterol but no coronary artery blockage, they aren't prescribed a bypass. If people have high cholesterol and later stage coronary artery disease including significant blockage of their coronary arteries, bypass surgery may reduce the risk of fatal heart attacks, but the "later stage" group with blocked coronary arteries will still have a higher death rate than the "early stage" group with no coronary artery blockage because they were sicker to begin with. The same is true for HIV, ARVs, and clinical AIDS and the death rate from it.
Example: Type 2 Diabetes
Having low blood sugar can be a symptom of type 2 diabetes, which can
lead to problems like diabetic shock, blindness, leg infections, and
death. People with mild type 2 diabetes may be able to satisfactorily
control it through diet and exercise. People with more serious,
later-stage type 2 diabetes cannot control it through diet and exercise
alone and may require insulin injections.If you were trying to understand whether insulin injections are helpful or harmful for low blood sugar due to type 2 diabetes, you might look at all your friends who have ow blood sugar due to type 2 diabetes and compare "people with low blood sugar due to type 2 diabetes who take insulin injections" with "people with low blood sugar due to type 2 diabetes who don't take insulin injections." Those who take insulin injections would have some number of deaths. Those who don't take insulin injections would have fewer deaths. Does it follow therefore that taking insulin injections as prescribed by a doctor increases your risk of death? This is precisely the (incorrect) conclusion you'd reach if you took this approach.
But no, that's not what it means. It means that if people have mild type 2 diabetes that's controllable via diet and exercise, they aren't prescribed insulin injections. If people have serious type 2 diabetes that can't be controlled through diet and exercise, insulin injections may reduce the risk of death, but the "later stage" group with serious type 2 diabetes will still have a higher death rate than the "early stage" group with no mild diabetes because they were sicker to begin with. The same is true for HIV, ARVs, and clinical AIDS and the death rate from it.
Valid and Invalid Ways of Studying the Effectiveness of ARVs for
HIV, Bypass Surgery for Coronary Artery Disease, and Insulin for Type 2
Diabetes
Here is a table summing up valid and invalid ways of studying the
effectiveness of ARVs, bypass surgery, and insulin and the correct or
incorrect conclusions you'll reach if you apply them.| Disease |
HIV Disease |
Coronary Artery Disease |
Type 2 Diabetes |
| Underlying Cause of Disease |
being infected with HIV (HIV
positive) |
body biochemistry is leading to
a high level of "bad" LDL cholesterol in the bloodstream |
body develops insulin
resistence, so sugar builds up to high levels in the bloodstream |
| "Early Stage" Group |
people who are HIV positive but
have a CD4 count greater than 350 and who have not experienced an
opportunistic infection |
people who have a tendency
towards high "bad" LDL cholesterol but can keep it down through diet,
exercise, and/or the use of statins like Lipitor |
people tend to have high blood
sugar levels but can control it through diet and exercise |
| "Late Stage" Group |
people who are HIV positive but have a CD4 count below 350 and/or have experienced an opportunistic infection | people who have developed
plaques that almost completely obstruct the flow of blood through the
coronary arteries to the heart |
people whose blood sugar levels
cannot adequately be controlled through diet and exercise |
| Partial List of Symptoms
Possibly Experienced by "Late Stage" Group |
CD4 count below 350, progression
to clinical AIDS (diagnosed when an HIV positive patient has a CD4
count under 200 or experiences an opportunistic infection), higher risk
of opportunistic infections, higher risk of death |
chest pain during exercise,
higher risk of heart attack, higher risk of death |
higher risk of diabetic
shock, blindness, infections in the legs, and death |
| Treatment That May Be
Prescribed to "Late Stage" Group |
use of antiretroviral
medications (ARVs) |
bypass surgery |
insulin injections |
| Scientific Question |
Does the use of ARVs for people
with "Late Stage" HIV disease increase the CD4 count, reduce
opportunistic infections, and reduce the death rate? |
Does bypass surgery for people with "Late Stage" coronary artery disease reduce the risk of heart attacks reduce the death rate? | Do insulin injections for people
with "Late Stage" diabetes reduce the risk of diabetic shock,
blindness, leg infections, and death? |
| Invalid Study Methodology for
Resolving the Scientific Question |
"Let's take all the HIV positive
people and compare the death rate of people who are on ARVs with the
death rate of people who aren't on ARVs." |
"Let's take all the people with high "bad" LDL cholesterol and compare the death rate of people who have bypass surgery with the death rate of people who don't have bypass surgery." | "Let's take all the people with high blood sugar levels due to type 2 diabetes and compare the death rate of people who take insulin injections with the death rate of people who don't take insulin injections." |
| Proposed "Test Group" in Invalid
Methodology |
HIV positive people who have
been prescribed and are taking ARVs |
people with high "bad" LDL
cholesterol who meet the criteria for and undergo heart bypass surgery |
people with high blood sugar
levels due to type 2 diabetes who have been prescribed and are taking
insulin injections |
| Proposed "Control Group" in
Invalid Methodology |
HIV positive people who do not
meet the criteria to be prescribed ARVs |
people with high "bad" LDL cholesterol who do not meet the criteria for bypass surgery | people with high blood sugar
levels due to type 2 diabetes who do not meet the criteria to be
prescribed insulin injections |
| Test Criterion |
death rate |
death rate | death rate |
| Invalid Conclusion Reached by
Invalid Study Methodology |
"HIV positive people who are on
ARVs have a higher death rate than HIV positive people not taking ARVs,
therefore ARVs must be harmful." (This is wrong.) |
"People with high "bad" LDL cholesterol who have bypass surgery have a higher death rate than people with high "bad" LDL cholesterol who don't have bypass surgery, therefore bypass surgery must be harmful." (This is wrong.) | "People with high blood sugar levels due to type 2 diabetes who take insulin injections have a higher death rate than people with high blood sugar levels due to type 2 diabetes who don't take insulin injections, therefore insulin injections must be harmful." (This is wrong.) |
| Fatal Flaw of Invalid Study
Methodology |
Error in test/control group
definition in study design. "Test Group" and "Control Group" are not
equivalent. "Test Group" members have CD4 counts under 350 and/or have
experienced an opportunistic infection. "Control Group" members have
CD4 counts above 350 and have not experienced an opportunistic
infection. Therefore, "Test Group" members are by definition sicker
(further progressed in HIV disease) than "Control Group" members and
are automatically expected to have a higher death rate. |
Error in test/control group definition in study design. "Test Group" and "Control Group" are not equivalent. "Test Group" members have advanced coronary artery disease with significant blockage of their coronary arteries. "Control Group" members have high "bad" LDL cholesterol but mild or no coronary artery blockage. Therefore, "Test Group" members are by definition sicker (further progressed in coronary artery disease) than "Control Group" members and are automatically expected to have a higher death rate. | Error in test/control group definition in study design. "Test Group" and "Control Group" are not equivalent. "Test Group" members have serious type 2 diabetes that cannot be controlled with diet and exercise. "Control Group" members have mild type 2 diabetes that can be satisfactorily controlled through diet and exercise. Therefore, "Test Group" members are by definition sicker than "Control Group" members and are automatically expected to have a higher death rate. |
| Valid Study Methodology for
Resolving the Scientific Question (Simplified for Illustrative Purposes
-- see note below) |
"Let's take the HIV positive
people who meet the criteria for prescribing ARVs, divide them into two
groups, give one group ARVs and one group a placebo, and compare the
death rate of the two groups." (see note below) |
"Let's take the people with high "bad" LDL cholesterol who meet the criteria for bypass surgery, divide them into two groups, give one group bypass surgery and one group a placebo pill, and compare the death rate of the two groups." (see note below) | "Let's take the people with high blood sugar levels due to type 2 diabetes who meet the criteria for prescribing insulin injections, divide them into two groups, give one group insulin injections and one group a placebo pill, and compare the death rate of the two groups." (see note below) |
| "Test Group" in Valid Methodology | HIV positive people who meet the
criteria for prescribing ARVs and are given ARVs |
people with high "bad" LDL
cholesterol who meet the criteria for bypass surgery and are given
bypass surgery |
people with high blood sugar levels due to type 2 diabetes who meet the criteria for insulin injections and are given insulin injections |
| "Control Group" in Valid Methodology | HIV positive people who meet the
criteria for prescribing ARVs and are given a placebo |
people with high "bad" LDL
cholesterol who meet the criteria for bypass surgery and are given a
placebo pill |
people with high blood sugar
levels due to type 2 diabetes who meet the criteria for insulin
injections and are given a placebo pill |
| Valid Conclusion Reached by Valid Study Methodology | For HIV positive people who meet
the criteria for prescribing ARVs, those who take ARVs have a lower
death rate than matched control subjects who don't take ARVs. Therefore
ARVs are effective for the treatment of "Late Stage" HIV disease. |
For people with high "bad" LDL cholesterol who meet the criteria for bypass surgery, those who undergo bypass surgery have a lower death rate than matched control subjects who don't undergo bypass surgery. Therefore bypass surgery is effective for the treatment of "Late Stage" coronary artery disease. | For people with high blood sugar levels due to type 2 diabetes who meet the criteria for insulin injections, those who take insulin injections have a lower death rate than matched control subjects who don't take insulin injections. Therefore insulin injections are effective for the treatment of "Late Stage" type 2 diabetes. |
Simplifications in Above Explanations
This page is by no means intended to give a complete explanation of the
causes, course, and/or treatment of HIV/AIDS, coronary
artery disease,
and type
2 diabetes. Coronary artery disease and diabetes in particular can
have multiple contributing factors. This page presents a simplified
explanation of HIV, coronary artery disease, and type 2 diabetes that
is for illustrative purposes only.The above "scientific study designs" are simplified for illustrative purposes. Real-world clinical studies would have to be more complex and would take more issues into consideration:
- Patient Safety and Ethical Study Design
- It was only during the early stages of the HIV epidemic that it was ethical to conduct a clinical study in which one group took a single ARV and the control group took a placebo. When AZT was the only proposed antiretroviral drug available and its safety and its effectiveness was yet unproven, this study methodology was in fact used for a scientific study called BW002, and it showed that only 1 of 145 patients treated with AZT died vs. 19 of 137 on the placebo; 24 on AZT had opportunistic infections vs. 45 on the placebo. [NIAID citing Fischl et al, "The efficacy of azidothymidine (AZT)...." N Engl J Med. 1987 Jul 23;317(4):185-91.] It was also used for the ACTG016 trial, which showed that HIV+ patients with CD4 counts of 200-500 on AZT were less likely to experience disease progression. [NIAID citing Fischl et al, "The Safety ..." Ann Intern Med. 1990 May 15;112(10):727-37.] Today, since the effectiveness of ARVs for treating HIV infection has been proven, it would not be ethical to conduct an experiment in which control subjects were given no ARVs. Instead, the test methodology would likely compare patients who were on two proven antiretrovirals plus a new unproven retroviral with patients who were on three proven antiretrovirals.
- Similarly, it is only when insulin first became available and its effectiveness for treating later stage diabetes was unproven that a study would likely have compared using insulin to not using insulin.
- Controlling for Additional Factors
- You would need to make sure that the overall health of the two
groups (test group and control group) was equivalent in other areas
such as smoking/nonsmoking, illicit drug use, etc., since if the two
groups were different in these areas, it could bias the study result
between the two groups.