Anda di halaman 1dari 4

What have we learned about randomized, controlled trials in

neonatal sepsis?
William O. Tarnow-Mordi, BA (Cantab), MBChB DCH, MRCP(UK), FRCPCH

Objective: To identify strategies to enable randomized, con- substantially greater sample sizes are needed to limit the play of
trolled trials in neonatal sepsis to recognize therapies that in- chance. Prospective meta-analysis using individual patient data
crease disability-free survival. is a promising strategy. It requires researchers to obtain funding
Methods: Rapid literature review. from their national agencies to conduct similar trials according to
Results: Randomized, controlled trials are the gold standard an agreed protocol with prespecified hypotheses, interventions,
for testing a therapy because they minimize bias. However, ran- power calculations, data sets, and measures of outcome. Pro-
domized, controlled trials must also minimize random error if they spective meta-analysis combines the methodologic advantages of
are to detect moderate (and realistic) improvements in rates of a single “megatrial” with the practical advantages that financial
disability-free survival reliably. This requires surprisingly large burdens are spread internationally and different national funding
samples (i.e., thousands rather than hundreds). Against this per- cycles can be more flexibly accommodated.
spective, most neonatal trials have been too small to be conclu- Conclusions: Prospective meta-analysis using individual pa-
sive, so most neonatal therapies remain incompletely evaluated. tient data is a promising strategy for achieving large-scale, ran-
As in specialties like cardiology and obstetrics, achieving reliable domized evidence in neonatal sepsis. (Pediatr Crit Care Med 2005;
trials in neonatal sepsis will require international collaboration, 6[Suppl.]:S146 –S149)
simpler data sets, more cost-effective recruitment strategies, less KEY WORDS: randomized, controlled trials; neonatal sepsis; pro-
exclusive criteria for selecting collaborators and patients, and an spective meta-analysis
appreciation by clinicians and data-monitoring committees that

R andomized, controlled trials For example, statisticians at a recent con- ment effects. As a result, many modest
in pediatric and neonatal sep- sensus conference on clinical trials in but important treatment effects have
sis are too few and too small. adult sepsis pointed out that if investiga- probably been missed (6, 7), and most
The need for more random- tors wanted to detect a moderate risk neonatal therapies remain incompletely
ized, controlled trials and meta-analyses difference of 5% (e.g., from 40% to 35%) evaluated.
is well recognized in adult, pediatric, and reliably when control event rates are
perinatal medicine (1–5). Both ap- around 30%– 40%, adequately powered
proaches avoid bias, if appropriately ana- Large, Simple Trials Are
studies would need around 4,000 patients
lyzed. Yet, if we want reliable answers Necessary and Feasible
(1) (Table 1).
about whether therapies decrease mortal- Many authors “resolve” this problem Yusuf et al. (8) gave four reasons for
ity and disability, avoiding bias is not by designing smaller trials, with samples undertaking large, simple trials to answer
enough. The effects of most treatments of a few hundred patients or fewer. These certain important questions. First, treat-
are likely to be moderate. To avoid being trials have only enough power to detect ments with big effects on mortality are
seriously misled by the play of chance either unrealistically large differences rare, so looking for moderate effects on
when looking for moderate effects, trials (e.g., from 30% to 15%) in categorical mortality in common conditions is im-
must also minimize random error, and primary outcomes, like death or disabil- portant. This requires really large trials.
this requires surprisingly large numbers. ity, or to demonstrate moderate differ- Second, because death is simple to mea-
ences in a continuous primary outcome, sure, trials with mortality as the main
like blood pressure or plasma bilirubin outcome can be large, with simple fol-
From the Department of Neonatal Medicine, Uni- (6). In a recent survey of systematic re- low-up. Third, treatments used widely by
versity of Sydney, Westmead Hospital, and The Chil-
dren’s Hospital at Westmead, New South Wales, Aus-
views of neonatal trials in the Cochrane busy clinicians must be easy to adminis-
tralia. Library, there were only 90 reviews of ter, so really large trials can have very
This work was supported by the Mannion Family trials with a categorical primary outcome simple regimens. Lastly, quantitative dif-
Fund—Center for the Critically Ill Child, Division of (such as mortality or survival). Among ferences in the degree of effect between
Critical Care Medicine at Children’s Hospital Boston,
these, the median number of outcomes subgroups are common. However, unex-
the PALISI Network, and the ISF.
Copyright © 2005 by the Society of Critical Care per review was only 54, with a range from pected qualitative differences in the di-
Medicine and the World Federation of Pediatric Inten- 1 to 1,284 (7). Clearly, most neonatal rection of effect—so that treatment ben-
sive and Critical Care Societies trials are too small to provide reliable efits some subgroups but harms others—
DOI: 10.1097/01.PCC.0000161576.09029.56 evidence about any but the largest treat- are unlikely. Therefore really large trials

S146 Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)


Table 1. Power calculations to detect moderate or large differences in disability-free survival with 90% creased the risk of necrotizing enteroco-
power at 2 (p ⫽ .05) litis (20).
Key features of all these trials were:
Adverse Outcome
defining an important clinical question,
Adverse Outcome in Treatment Absolute Risk Difference Total Sample Needed
in Controls, % Group, % (Control Minus Treatment), % (Both Groups Combined) ensuring that the eligibility criteria and
data collection forms were as simple as
15 13 2 12,846 possible, and reducing extra work for
40 35 5 4,016 busy participants to a minimum. No re-
35 30 5 3,764 strictions were placed on collaboration,
30 26 4 5,392
25 21 4 4,748
which was open to any center wishing to
20 15 5 2,502 join. The costs of participation were min-
30 20 10 824 imal, and clinicians received little or no
30 15 15 348 reimbursement. Their prime motivation
was a desire to help to answer the ques-
In this example, risk differences of 2% to 5% would be considered moderate but realistic and
tion and thus improve outcomes for their
worthwhile effects of treatment; risk differences of ⱖ10% are considered unrealistically large and
rarely seen in the “real world.”
patients.

Achieving Larger Trials in Small


do not need to collect extensive data or to oral mononitrate vs. placebo, and intra-
Subspecialties: The Example of
define many subgroups of patients. Also, venous magnesium sulfate vs. open con- Acute Lymphoblastic Leukemia
because having many subgroups of pa- trol. It recruited 58,050 patients in Cardiology, stroke, adult cancer,
tients is unlikely to enhance the primary ⬎1,000 hospitals in ⬍3 yrs. It showed trauma, and obstetrics have high vol-
analyses, really large trials can have sim- lower mortality with captopril but no umes of patients. For example, there are
ple plans of analysis. These authors also benefits from mononitrate or magnesium about 1.5 million myocardial infarctions,
recommended factorial (e.g., 2 ⫻ 2) trial sulfate. 90,000 infants born weighing ⬍1500 g,
designs to test more than one therapy The International Stroke Trial (13) and 2,400 children with acute lympho-
simultaneously for the price of one. The and Chinese Acute Stroke Trial (14) each blastic leukemia in the United States an-
presence of interactions between treat- recruited about 20,000 patients. Half re- nually. Small specialties therefore need
ments is not a contraindication to a fac- ceived aspirin within 48 hrs of a stroke, even greater collaboration. Pediatric on-
torial design. Even if a qualitative inter- and aspirin was continued for up to 4 cologists have integrated randomized,
action exists, a factorial trial may be the wks. A prospectively planned meta- controlled trials into routine clinical
design of choice, as it alone will give an analysis of the individual data from these practice with impressive results. In the
unbiased estimate of the truth (8). 40,000 patients confirmed that aspirin re- 1960s, acute lymphoblastic leukemia had
duces recurrent stroke and mortality in a a case fatality rate of ⬎95%. After earlier
Simple Trials and Large-Scale, wide range of patients (15). Aspirin trials based on protocols of the United
Randomized Evidence in Other quickly became standard therapy. States Children’s Cancer Study Group,
The MAGPIE trial (17) recruited the Tenth Medical Research Council
Specialties
10,141 women at risk of eclampsia in 33 Acute Lymphoblastic Leukaemia trial en-
Many large, simple, and efficient trials countries. It showed that magnesium sul- rolled 1,612 children between 1985 and
have been conducted in cardiology, fate halved the risk of eclampsia, from 1990. This represented ⬎90% of children
stroke, trauma cancer, and obstetrics 1.9% in the placebo group to 0.8% in the with acute lymphoblastic leukemia in the
with assistance from the Clinical Trial treated group. UK (21). Five-year survival in acute lym-
Service Unit, Oxford, UK (9 –18). Within The ORACLE trials recruited 4,826 phoblastic leukemia is now 85%. How-
10 yrs, the International Study of Infarct women with preterm prelabor rupture of ever, as mortality and complications de-
Survival “megatrials” reduced mortality fetal membranes (trial 1) (18) and 6,295 crease, future improvements will require
after myocardial infarction from around women with spontaneous preterm labor larger trials and even greater interna-
13% in controls to 7% in the best treat- with intact membranes (trial 2) (19) in a tional collaboration (22).
ment arm (9 –12). Each had substantial placebo-controlled, 2 ⫻ 2, factorial com-
effect on clinical practice, especially the parison of co-amoxiclav, erythromycin, Misleading Interim Evidence of
International Study of Infarct Survival-2 both, or neither. There was no reduction Futility, Harm, or Benefit:
(10). This was a placebo-controlled, fac- in the composite outcome of neonatal
Pitfalls in Data Monitoring
torial, 2 ⫻ 2 study comparing aspirin, mortality, chronic lung disease, or cere-
streptokinase, both, or neither in 17,187 bral abnormality on ultrasound. Impor- Clinicians and members of data and
patients. It showed that aspirin and strep- tantly, in a systematic review of 14 ran- safety monitoring committees need to be
tokinase each reduced the relative risk of domized, controlled trials in 6,559 aware that accumulated data are fre-
mortality by about 20%–25% and to- women with preterm prelabor rupture of quently misleading, owing to the highly
gether reduced mortality by about 40%. fetal membranes, prenatal antibiotics unpredictable effects of chance. In a trial
International Study of Infarct Survival-4 prolonged pregnancy and reduced mater- of intravenous streptokinase after myo-
(12) was a 2 ⫻ 2 ⫻ 2, controlled, factorial nal infection, neonatal infection, use of cardial infarction, GISSI-1, there was no
study of captopril (an angiotensin- surfactant, oxygen therapy, and abnormal clear interim evidence of benefit at the
converting enzyme inhibitor) vs. placebo, cerebral ultrasound. Co-amoxiclav in- half way point, despite recruiting ⬎5,000

Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.) S147


patients with ⬎600 deaths (23). However, patients within trials, 3) outcome defini- trials in pediatric critical care: Rarely done
when the final sample size of 11,712 was tions (primary and secondary), 4) pro- but desperately needed. Pediatr Crit Care
reached, there was strong evidence that posed subgroups, and 5) plan of analysis. Med 2002; 3:102–106
streptokinase saved lives (mortality rate Trials are prospective registered, in- 3. Crowley P, Chalmers I, Keirse MJ: The effects
of corticosteroid administration before pre-
of 10.7% in treated patients vs. 13% in cluding any trials with factorial designs
term delivery: An overview of the evidence
controls, p ⫽ .0002) (24). Premature ter- that incorporate other interventions.
from controlled trials. Br J Obstet Gynaecol
mination would have contributed to con- Each trial will collect and contribute in- 1990; 97:11–25
tinuing uncertainty. dividual patient data. Individual trials will 4. Peto R, Baigent C: Trials: The next 50 years.
In a trial of the effects of the angio- all be protected by being published before Large scale randomised evidence of moderate
tensin-converting enzyme inhibitor the joint publication of the PMA. New benefits. BMJ 1998; 317:1170 –1171
benazepril in slowing nephropathy in 583 questions or trials may be added to the 5. Pogue J, Yusuf S: Overcoming the limita-
renal patients, there was a nominally sig- PMA later, but only if all trial data are still tions of current meta-analysis of randomised
nificant increase in mortality (p ⫽ .04) blinded. The advantages of a PMA are that controlled trials. Lancet 1998; 351:47–52
(25). However, a systematic review of 20 it provides the same strengths and bene- 6. Zhang B, Schmidt B: Do we measure the
trials of angiotensin-converting enzyme fits as a single megatrial and greater right end points? A systematic review of pri-
inhibitors in 2,383 patients with diabetic power and more reliable estimates of mary outcomes in recent neonatal random-
nephropathy showed a 21% reduction in treatment effects for less frequent, clini- ized clinical trials. J Pediatr 2001; 138:76 – 80
7. Sinclair JC, Haughton DE, Bracken MB, et al:
relative risk of all-cause mortality (p ⫽ cally important outcomes. It allows
Cochrane neonatal systematic reviews: A sur-
.04) (26). The adverse mortality in the greater power than in individual trials to vey of the evidence for neonatal therapies.
smaller trial probably reflects the play of evaluate events in important subgroups Clin Perinatol 2003; 30:285–304
chance. and permits regression and time-series 8. Yusuf S, Collins R, Peto R: Why do we need
In a multiple-center trial of the tissue analyses for patients within and across some large, simple randomized trials? Stat
factor pathway inhibitor, tifacogin, in individual trials. It also allows for pre- Med 1984; 3:409 – 422
adult sepsis, an interim analysis, after 772 specified differences in specific aspects of 9. Randomised trial of intravenous atenolol
patients, showed a mortality rate of individual trial protocols, for example, among 16,027 cases of suspected acute myo-
38.9% in the placebo group vs. 29.1% in the age at which disability is to be deter- cardial infarction: ISIS-1. First International
the treated patients (p ⫽ .006). Despite mined. PMA distributes the cost of a large Study of Infarct Survival Collaborative
this apparent treatment benefit, the trial among research agencies in individ- Group. Lancet 1986; 2:57– 66
study’s data safety and monitoring com- ual countries, in contrast to funding by a 10. Randomised trial of intravenous streptoki-
nase, oral aspirin, both, or neither among
mittee decided against stopping the trial single agency. Thus, centers in collabo-
17,187 cases of suspected acute myocardial
early. At its completion, with 1,754 pa- rating countries would be sponsored by
infarction: ISIS-2. ISIS-2 (Second Interna-
tients, the mortality rates were 33.9% in their own research agency and would tional Study of Infarct Survival) Collabora-
the placebo group vs. 34.2% in the conduct their own trial. The PMA would tive Group. Lancet 1988; 2:349 –360
treated patients (p ⫽ .88) (27). be registered and approved by the Co- 11. ISIS-3: A randomised comparison of strep-
In summary, randomized evidence chrane Prospective Analysis Methods tokinase vs tissue plasminogen activator vs
from smaller samples can both underes- Group. Overall, a PMA provides greater, anistreplase and of aspirin plus heparin vs
timate and overestimate the treatment pragmatic flexibility in achieving the ob- aspirin alone among 41,299 cases of sus-
effects seen in larger samples. Both the jectives of a single megatrial. pected acute myocardial infarction. ISIS-3
observed effect and the 95% confidence (Third International Study of Infarct Sur-
interval are biased upward when random- vival) Collaborative Group. Lancet 1992; 339:
Conclusion
ization stops early because the tendency 753–770
Although randomization avoids bias, 12. ISIS-4: A randomised factorial trial assessing
is to stop when the observed treatment
it is less well recognized that random- early oral captopril, oral mononitrate, and
difference is on a “random high,” greater
ized, controlled trials must minimize intravenous magnesium sulphate in 58,050
than the underlying truth (28). patients with suspected acute myocardial in-
random error if moderate effects are to be
farction. ISIS-4 (Fourth International Study
detected reliably. This requires surpris-
Prospective Meta-analysis and of Infarct Survival) Collaborative Group.
ingly large numbers (i.e., thousands,
Large-Scale, Randomized Lancet. 1995; 345:669 – 685
rather than hundreds) of patients. Such 13. The International Stroke Trial (IST): A ran-
Evidence in Neonatal Sepsis trials will require international collabora- domised trial of aspirin, subcutaneous hepa-
One promising strategy for overcom- tion and considerably simpler data sets. rin, both, or neither among 19435 patients
ing the challenge of large sample sizes in Prospective meta-analyses using individ- with acute ischaemic stroke. International
small specialties like neonatology or pe- ual patient data are a promising strategy Stroke Trial Collaborative Group. Lancet
diatric intensive care is prospective pool- for achieving large-scale, randomized ev- 1997; 349:1569 –1581
ing of individual patient data from two or idence. 14. CAST: Randomised placebo-controlled trial
of early aspirin use in 20,000 patients with
several trials addressing the same ques-
acute ischaemic stroke. CAST (Chinese Acute
tion (5, 29, 30). REFERENCES
Stroke Trial) Collaborative Group. Lancet
The key features of a prospective 1997; 349:1641–1649
1. Cohen J, Guyatt G, Bernard GR, et al: New
meta-analysis (PMA) are that, before any strategies for clinical trials in patients with 15. Chen ZM, Sandercock P, Pan HC, et al: In-
results of individual trials are known, col- sepsis and septic shock: UK Medical Research dications for early aspirin use in acute isch-
laborators agree to a formal protocol that Council International Working Party. Crit emic stroke: A combined analysis of 40,000
includes 1) specific hypotheses/objec- Care Med 2001; 29:880 – 886 randomized patients from the Chinese Acute
tives, 2) eligibility criteria of trials and 2. Randolph AG, Lacroix J: Randomized clinical Stroke Trial and the International Stroke

S148 Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.)


Trial. On behalf of the CAST and IST collab- systematic review. Obstet Gynecol 2004; 104: 26. Strippoli GF, Craig M, Deeks JJ, et al: Effects
orative groups. Stroke 2000; 31:1240 –1249 1051–1057 of angiotensin converting enzyme inhibitors
16. Roberts I, Yates D, Sandercock P, et al: Effect of 21. Chessells JM, Bailey C, Richards SM: Intensifi- and angiotensin II receptor antagonists on
intravenous corticosteroids on death within 14 cation of treatment and survival in all children mortality and renal outcomes in diabetic ne-
days in 10,008 adults with clinically significant with lymphoblastic leukaemia: Results of UK phropathy: Systematic review. BMJ 2004;
head injury (MRC CRASH trial): Randomized Medical Research Council trial UKALL X. Med- 329:828
placebo-controlled trial. CRASH trial collabo- ical Research Council Working Party on Child- 27. Abraham E, Reinhart K, Opal S, et al: Effi-
rators. Lancet 2004; 364:1321–1328 hood Leukaemia. Lancet 1995; 345:143–148 cacy and safety of tifacogin (recombinant
17. Altman D, Carroli G, Duley L, et al: Do 22. Vora A: Acute lymphoblastic leukemia: Opti- tissue factor pathway inhibitor) in severe
women with pre-eclampsia, and their babies, mizing treatment strategies in children. Pae- sepsis: A randomized controlled trial. OP-
diatr Drugs 2002; 4:405– 416
benefit from magnesium sulphate? The Mag- TIMIST Trial Study Group. JAMA 2003;
23. Yusuf S: Challenges in the conduct and in-
pie Trial: A randomised placebo-controlled 290:238 –247
terpretation of phase II (pilot) randomized
trial. Lancet 2002; 359:1877–1890 28. Pocock S, White I: Trials stopped early: Too
trials. Am Heart J 2000; 139:S136 –S142
18. Kenyon SL, Taylor DJ, Tarnow-Mordi W: good to be true? Lancet 1999; 353:943–944
24. Effectiveness of intravenous thrombolytic
Broad-spectrum antibiotics for preterm, pre- 29. Simes RJ: Prospective meta-analysis of cho-
treatment in acute myocardial infarction:
labour rupture of fetal membranes: The OR- Gruppo Italiano per lo Studio della Strep- lesterol-lowering studies: The Prospective
ACLE I randomised trial. ORACLE Collabo- tochinasi nell’Infarto Miocardico (GISSI). Pravastatin Pooling (PPP) Project and the
rative Group. Lancet 2001; 357:979 –988 Lancet 1986; 1:397– 402 Cholesterol Treatment Trialists (CTT) Col-
19. Kenyon SL, Taylor DJ, Tarnow-Mordi W: 25. Maschio G, Alberti D, Janin G, et al: Effect of laboration. J Cardiol 1995; 76:122C–126C
Broad-spectrum antibiotics for spontaneous the angiotensin-converting-enzyme inhibi- 30. Simes J, Furberg CD, Braunwald E, et al:
preterm labour: The ORACLE II randomised tor benazepril on the progression of chronic Effects of pravastatin on mortality in patients
trial. ORACLE Collaborative Group. Lancet renal insufficiency: The Angiotensin-Con- with and without coronary heart disease
2001; 357:989 –994 verting-Enzyme Inhibition in Progressive across a broad range of cholesterol levels:
20. Kenyon S, Boulvain M, Neilson J: Antibiotics Renal Insufficiency Study Group. N Engl The Prospective Pravastatin Pooling project.
for preterm rupture of the membranes: A J Med 1996; 334:939 –945 Eur Heart J 2002; 23:207–215

Pediatr Crit Care Med 2005 Vol. 6, No. 3 (Suppl.) S149

Anda mungkin juga menyukai