How to apply results from randomized trials and systematic

Nephrology 15 (2010) 277–280
C l i n i c a l R e s e a rc h f o r N e p h ro l o g i s t s
How to apply results from randomized trials and systematic
reviews to individual patient care
nep_
1,2
ELISABETH M HODSON
277..280
1,2
and JONATHAN C CRAIG
1
Cochrane Renal Group, Centre for Kidney Research, The Children’s Hospital at Westmead, and 2School of Public Health, University of Sydney, Sydney, New
South Wales, Australia
KEY WORDS:
data application, individual patient care
Correspondence:
Dr Elisabeth Hodson, Centre for Kidney
Research, The Children’s Hospital at
Westmead, Locked Bag 4001, Westmead, NSW
2145, Australia. Email: [email protected]
Accepted for publication 1 November 2009.
Accepted manuscript online 25 January 2010.
doi:10.1111/j.1440-1797.2010.01281.x
SUMMARY AT A GLANCE
Using results from randomized controlled
trials and population-based cohort studies,
a clinician may apply the results to the
management of an individual patient by
answering five questions to determine
whether the patient will achieve more
benefit than harm from the intervention.
ABSTRACT:
A clinician may apply the results from randomized controlled trials and
population-based cohort studies to the management of an individual
patient to determine whether the patient will achieve more benefit than
harm from the intervention. From the data the clinician should determine
what are the benefits and harms of the intervention, whether there are any
variations in the relative treatment effect, whether the treatment effect
varies with different baseline risks of disease in untreated patients, what are
the predicted reductions in absolute risk of disease for individuals and
whether the benefits outweigh the risks for their patient. If the patient is at
a low risk of the outcome, the harms of therapy may not justify its use to
prevent or treat the disease. However, if the patient is at a high risk of
developing the outcome, he or she is likely to gain more benefit than harm
from the therapy.
Your 52-year-old patient is to receive a living-donor renal
transplant from his son. Both father and son are cytomegalovirus (CMV) IgG positive. Should the recipient receive
prophylaxis with valganciclovir to prevent CMV disease? You
identify a systematic review1,2 (level I evidence3) of randomized controlled trials (RCT) comparing antiviral medications
(ganciclovir, valaciclovir, aciclovir) with placebo or no specific treatment and a single large well-designed RCT (level II
evidence3) comparing valganciclovir with ganciclovir4 to
answer this question.
How should you use the results of these studies in your
decision? Frequently, readers of trials and systematic reviews
consider the eligibility criteria for individual studies in deciding whether the results are applicable to their patient, but
this is inappropriate for two reasons. Typically triallists decide
upon eligibility criteria for reasons of efficiency and not
applicability. They seek to restrict their trial to patients who
are at high risk of the outcome that the intervention is
designed to prevent, at low risk of competing events and at
low risk of adverse events. This often means the exclusion of
© 2010 The Authors
Journal compilation © 2010 Asian Pacific Society of Nephrology
children, elderly people and those with comorbidities. These
strategies mean that the sample size can be smaller than if a
broader patient population is included. It does not mean that
the intervention will not be beneficial in the excluded group
of patients, many of whom are at highest risk. Furthermore,
most patients included in trials, even placebo-controlled
trials, do not develop the outcome of interest and so could
not benefit from the intervention. Using the trial eligibility
criteria, as the sole basis for using an intervention, will lead
to the overtreatment of many patients and could subject
them to unnecessary harm.
What is an alternative approach for applying research
data to individual patient care? Developed by the Cochrane
Applicability Methods group and adopted by the National
Health and Medical Research Council, a five-step framework (Table 1) can be used. This importantly requires the
use of cohort including registry data as well as RCTs. In
short it identifies the group of patients in whom the likely
benefits exceed the potential harms.5 The questions to be
asked are:
277
EM Hodson and JC Craig
1 What are the benefits and harms of the intervention?
2 Are there variations in the relative treatment effect?
3 Does the treatment effect vary with different baseline risks
of disease in untreated patients?
4 What are the predicted reductions in absolute risk of
disease for individuals?
5 Do the benefits of the intervention outweigh the harms?
WHAT ARE THE BENEFITS AND HARMS OF
THE INTERVENTION?
This step requires you to list all of the important good and
bad outcomes known for the intervention. For your potential
renal transplant recipient, the main beneficial clinically relevant effect is prevention of CMV disease and the systematic
review1,2 reported a significant reduction in the risk of this
outcome (relative risk 0.42; 95% confidence intervals 0.34–
0.52) with antiviral medications compared with placebo or
no specific therapy. Additional outcomes to consider are
overall mortality, mortality from CMV disease, acute rejection, graft loss, other opportunistic infections, duration and
cost of hospitalization and treatment. Adverse effects of antiviral medications to consider include leucopenia and infection, renal toxicity and neurological side effects.
ARE THERE VARIATIONS IN THE RELATIVE
TREATMENT EFFECT?
This step requires you to consider whether the treatment
effect varies in relative terms, particularly with known
patient characteristics. In this scenario, does the relative
treatment effect vary across different patient groups (liver,
kidney, heart recipients), donor and recipient CMV status,
different types, dosages or durations of prophylactic therapy
or different immunosuppressive regimens? The systematic
review1,2 did not detect any significant variations in these
variables when explored by testing for interactions between
studies. Given these findings then it is reasonable to take the
overall treatment effect of antiviral medications in solid
organ transplant recipients rather than a treatment effect
from a specific subgroup of patients. There are many other
scenarios where the treatment effect varies predictably with
Table 1 Five steps used to determine applicability of results of systematic
reviews and randomized controlled trials to an individual patient
Five steps used to apply results to an individual patient
1 What are the benefits and harms of the intervention?
2 Are there variations in the relative treatment effect?
3 Does the treatment effect vary with different baseline risks of disease in
untreated patients?
4 What are the predicted reductions in absolute risk of disease for
individuals?
5 Do the benefits of the intervention outweigh the harms?
278
known patient or intervention features. For example, it is
well known that the relative treatment effect of antioestrogen drugs like tamoxifen varies with oestrogen receptor status so that the treatment effect from the relevant
subgroup should be applied rather than the overall effect.
DOES THE TREATMENT EFFECT VARY WITH
DIFFERENT BASELINE RISKS OF DISEASE IN
UNTREATED PATIENTS?
This step requires that you consider the baseline level of risk
for the outcome of interest, in this case CMV disease without
prophylaxis, and whether this outcome shows the expected
reduction with prophylaxis. In general, benefits of treatment
increase while rates of adverse effects remain unchanged as
the baseline risk of disease increases.6 Therefore low-risk
patients have less to gain from an intervention than high-risk
patients and may not gain sufficient benefit to outweigh the
adverse effects of treatment even though the relative benefit
is the same for both high and low-risk patients. For example,
as demonstrated in Table 2, a relative risk of 0.5 (a 50%
reduction in the risk of disease) could mean a reduction in
absolute risk of disease from a baseline risk of 20% to 10% (a
10% absolute risk reduction) or a reduction from a baseline
risk of 2% to 1% (a 1% absolute risk reduction). Accurate
information on the baseline risk of disease should be
obtained from large prospective cohort studies rather than
from RCTs, where participants may not be representative of
the general population.
WHAT ARE THE PREDICTED REDUCTIONS IN
ABSOLUTE RISK FOR INDIVIDUALS?
In this step you need to predict the likely reduction in absolute risk for your patient if he receives prophylaxis. The
Table 2 Changes in absolute risk difference with antiviral prophylaxis for the
same relative risk of disease at different baseline risks of disease
Antiviral
agent given
Antiviral agent
not given
CMV disease
10
20
No CMV disease
90
80
Total
100
100
Baseline risk of disease without treatment = 20%.
Relative risk: 10/100 ⫼ 20/100 = 0.5 (50% relative risk reduction).
Absolute risk difference: 20/100 - 10/100 = 0.1 (10% absolute risk reduction).
CMV disease
10
20
No CMV disease
990
980
Total
1000
1000
Baseline risk of disease without treatment = 2%.
Relative risk: 10/1000 ⫼ 20/1000 = 0.5 (50% relative risk reduction).
Absolute risk difference: 20/1000 - 10/1000 = 0.01 (1% absolute risk
reduction).
CMV, cytomegalovirus.
© 2010 The Authors
Journal compilation © 2010 Asian Pacific Society of Nephrology
Applying results to individual patients
Table 3 Benefits and harms of antiviral medications for preventing CMV disease with antiviral medications following renal transplant
CMV risk
Low risk
Moderate risk
High risk
Serostatus
D§/R+¶
D+/R-††
D+/R- & antibody therapy
No. with CMV disease per 100 patients
Without
prophylaxis†
With
prophylaxis‡
No.
prevented
7
28
59
3
12
25
4
16
39
No. with harm
per 100 patients
7
7
7
†Data from references 7 and 8.
‡Calculated from summary estimate of relative risk (0.42).
§CMV serostatus positive or negative.
¶CMV seropositive.
††CMV seronegative.
CMV, cytomegalovirus.
higher the risk your patient has of developing CMV disease,
the greater the benefit to him of CMV prophylaxis. For
example, CMV seronegative transplant recipients of CMV
seropositive kidneys are at a higher risk of CMV disease than
CMV seropositive recipients of CMV seropositive or negative
kidneys. However, all treated patients are at the same risk of
adverse effects of prophylaxis. From the systematic review,
the relative risk for CMV disease with prophylaxis was 0.42
or 42% so the relative risk reduction was 58%. Based on data
of baseline risk obtained from population-based cohort
studies, a 58% relative risk reduction could mean a reduction in absolute risk from 28% to 12% for CMV seronegative
recipients of CMV seropositive kidneys or from 7% to 4% for
CMV seropositive recipients (Table 3).
DO THE BENEFITS OF THE INTERVENTIONS
OUTWEIGH THE HARMS?
Finally you are now ready to decide if the benefits of CMV
prophylaxis for your CMV seropositive renal transplant
recipient with a CMV seropositive donor outweigh the
potential harms of the medications. Of 100 CMV seronegative recipients of CMV positive kidneys, 28 will develop CMV
disease.7,8 With antiviral medications, 12 recipients will
develop CMV disease so 16 will avoid CMV disease (Table 2).
Overall about eight suffer transitory neurological side effects
with valaciclovir, 11 suffer renal dysfunction with ganciclovir
and two have their prophylaxis ceased temporarily because
of adverse effects.1,2 In such patients it is clear that CMV
prophylaxis is indicated. In contrast your patient has a 7%
risk7,8 of CMV disease, which with CMV prophylaxis can be
expected to fall to 4%. The risk of adverse effects is
unchanged. These adverse effects are usually mild, while
CMV disease can result in significant morbidity for your
patient. Considering the overall condition of your patient, do
you think that the harms of the medications outweigh the
benefits in preventing CMV disease?
You wish to treat your patient with valganciclovir, which is
better absorbed than oral ganciclovir. Valganciclovir became
© 2010 The Authors
Journal compilation © 2010 Asian Pacific Society of Nephrology
available after trials had demonstrated the benefit of antiviral
medications so that a placebo-controlled RCT of valganciclovir was not considered ethical. Can the results of the systematic review be applied to valganciclovir? The RCT comparing
valganciclovir with ganciclovir4 found no significant differences in the risk of CMV disease and other outcomes including harms suggesting that the outcomes demonstrated in
placebo or no-treatment trials could be extrapolated to
valganciclovir.
Using information from systematic reviews and RCT on
the benefits and harms of therapy while obtaining baseline
risks for the outcome from population-based cohort studies,
you can individualize treatment for your patients. If your
patient is at a low risk of the outcome, the harms of therapy
may not justify its use to prevent the outcome of interest.
However, if your patient is at a high risk of developing the
outcome, he or she is likely to gain more benefit than harm
from the therapy.
CONCLUSIONS
1 Using a 5-point framework, results of RCT and systematic
reviews can be applied to individual patients.
2 The benefits of treatment increase as the baseline risk of
disease increases while adverse effects of therapy remain
unchanged.
3 Low-risk patients have less to gain from an intervention
than high-risk patients so benefits may not outweigh harms
though the relative benefit is the same for both high- and
low-risk patients.
REFERENCES
1. Hodson EM, Jones CA, Webster AC et al. Antiviral medications to
prevent cytomegalovirus disease and early death in recipients of
solid-organ transplants: A systematic review of randomised
controlled trials. Lancet 2005; 365: 2105–15.
2. Hodson EM, Craig JC, Strippoli GFM, Webster AC. Antiviral
medications for preventing cytomegalovirus disease in solid organ
transplant recipients. Cochrane Database Syst. Rev. 2008, Issue 2. Art.
No.: CD003774. DOI: 10.1002/14651858.CD003774.pub3.
279
EM Hodson and JC Craig
3. National Health and Medical Research Council. NHMRC Additional
Levels of Evidence and Grades for Recommendations for Developers of
Guidelines. Stage 2 Consultation: Early 2008 to End June 2009. Canberra:
Commonwealth of Australia, 2008.
4. Paya C, Humar A, Dominguez E et al. Efficacy and safety of
valganciclovir vs oral ganciclovir for prevention of cytomegalovirus
disease in solid organ transplant recipients. Am. J. Transplant. 2004;
4: 611–20.
5. National Health and Medical Research Council. How to Use the
Evidence: Assessment and Application of Scientific evidence. Canberra:
Commonwealth of Australia, 2000.
280
6. Glasziou PP, Irwig LM. An evidence based approach to
individualising treatment. Br. Med. J. 1995; 311: 1356–9.
7. Waiser J, Budde M, Schreiber M et al. Effectiveness of deferred
therapy with ganciclovir in renal allograft recipients with
cytomegalovirus disease. Transplant. Proc. 1998; 30: 2083–5.
8. Smith SR, Butterly DW, Conlon PJ, Harland RC, Emovon OE.
Incidence of cytomegalovirus in renal disease without
antilymphocyte induction: Is prophylaxis necessary? Transplant. Proc.
1998; 30: 2097–9.
© 2010 The Authors
Journal compilation © 2010 Asian Pacific Society of Nephrology