Probiotics reduce mortality

Should parents of eligible infants be offered routine probiotics?

William Odita Tarnow-Mordiwilliamtm@med.usyd.edu.au

WINNER Centre for Newborn Research, Westmead Hospital

NHMRC Clinical Trials Centre, University of Sydney

• This presentation will discuss unapproved or investigational use of various probiotic preparations.

• Professor Tarnow-Mordi has no commercial or financial relationship relating to this presentation.

The many researchers, clinicians and families who have contributed to RCTs or systematic reviews of probiotics in preterm infants deserve great credit.

Their work is sincerely acknowledged.

Outline• Three current priorities

• Evidence based medicine and patient preference

• Parallels with antibiotics for colorectal surgery

• Rationale for more placebo RCTs

• What should parents be told?

• Can we reach a consensus for global collaboration?

Three current priorities

Continue research on probiotics

Disclose to parents and Institutional Review Boards current evidence that probiotics halve mortality rates

Assess whether probiotics can be made more widely accessible in RCTs and cohort studies

Evidence based medicine and

patient preference

Evidence based medicine Sackett et al, BMJ 1996: 312: 71-72

decision making is based on three fundamentals:-

patients' circumstancesbest research evidencepatients' preferences

Physicians' and patients' choices in evidence based practice

“…the clinician must consider the patient's preferences and likely actions …”

Haynes RB, Deveraux PJ, Guyatt GH. BMJ 2002; 324: 1350.

Howard Bauchner, 2010 Editor: Journal Watch Pediatrics and Adolescent Medicine

“ My suggestion is to discuss with parents the benefits, as well as arguments against routine probiotic use, and let them decide.”http://pediatrics.jwatch.org May 26, 2010

Edmund Hey (1934 – 2009) Editor: Neonatal Formulary 5

“ Do we, knowing what we now know, have the right to deny parents the option of giving a probiotic ...?”

Hey, E. Comment: Neonatal Formulary 5Has the time come to start using

probiotics more widely?http://www.blackwellpublishing.com/

medicine/bmj/nnf5/pdfs/comment/prob_com_jul09.pdf

Who is setting the agenda?

Parents’ preferences and options have received little attention so far.

The James Lind Alliance… to bring patients and clinicians together in 'Priority Setting Partnerships' to identify and prioritise the unanswered questions that they agree are most important. http://www.lindalliance.org

A key questionAfter each death without access to open label probiotic - particularly deaths from NEC - will clinicians, IRBs, managers and parents be satisfied that sufficient information and choices were given?

Parallels with antibiotics for

colorectal surgery

Sir Iain Chalmers, 2007Editor: James Lind Library

“Would any of you have agreed to participate in a placebo controlled trial of prophylactic antibiotics for colorectal surgery after 1975?” http://www.crash2.lshtm.ac.uk/

Reduction of perioperative deaths by antibiotic prophylaxis for colorectal surgery

Antibiotic prophylaxis for colorectal surgeryLau et al: J Clin Epidemiol 1995

By 1975, 10 RCTs had been done in 603 patients

Cumulative meta analysis showed a 75% reduction in the odds of mortality

In the next 12 years, 11 more RCTs reported another 928 patients, with little change in the odds of death.

Half - over 450 - received no antibiotics.

On balance of probability, most deaths in these patients could be ascribed to that omission.

Antibiotic prophylaxis for colorectal surgery Lau et al: J Clin Epidemiol 1995

“Trials of individual antibiotics had too few patients to show a significant result.

If the question is ‘does antibacterial prophylaxis reduce mortality in colon surgery patients?’ pooling all antibiotic trials is entirely acceptable.”

Antibiotic prophylaxis for colorectal surgeryLau et al: J Clin Epidemiol 1995

“ Withholding all antibiotics from control group patients is hardly justifiable if it is known from pooling of all antibiotic trials that … lives are saved.”

Antibiotic prophylaxis for colorectal surgery Lau et al: J Clin Epidemiol 1995

“If the question is ‘which antibiotic regimen is best ...?’, placebo controlled trials will not provide a reliable answer.”

“For that, multiple treatment arm ‘head to head’ RCTs are needed.

Placebo groups without antibiotics will not be ethically acceptable.”

Antibiotic prophylaxis for colorectal surgery

Lau et al: J Clin Epidemiol 1995

Two important differences• Antibiotics were routinely available for

adults undergoing colorectal surgery and clinicians were experienced in using them.

• Previously evaluated, locally approved probiotics are not yet routinely accessible for preterm infants in most jurisdictions.

As probiotics become more widely available, parallels with antibiotic prophylaxis in colorectal surgery will become closer.

Rationale for further placebo RCTs

Rationale for further placebo RCTs1. To identify the optimum regimen - ?

2. To refute evidence of lower mortality or morbidity - ?

3. Better estimates of specific products’ efficacy - ?

4. Pooling RCTs of different products is unacceptable - ?

5. Better estimates of short and long term risk - ?

6. The current meta analysis shows heterogeneity - X

7. Access to a rigorously tested probiotic will assist clinicians to gain familiarity and experience - √

8. Many clinicians remain substantially uncertain - √

1. To identify the optimum regimen - ?

• 2 arm placebo RCTs cannot reliably identify the optimum probiotic regimen.

• This will require multiple arm RCTs

2. To refute evidence of lower mortality - ?

New RCTs are very unlikely to refute or nullify current evidence of lower mortality

What would it take to nullify the Relative Risk Reduction of 0.42 (0.29 – 0.62, p < 0.00001) for mortality in Deshpande’s systematic review?

It would need a change in Relative Risk of

≥ 1/ 0.42 = 2.2 or more

i.e. more than double

This would require one of 2 scenarios:-

(A) ~ 2,000 new patients in RCTs showing the opposite effect on mortality, or

(B) ~ 4,500 new patients in RCTs with RR = 1.0

Given the substantial prior evidence from

(i) cohort studies (ii) animal physiology (iii) bench top research

scenarios (A) and (B) are very unlikely.

Hoyos AB. 55% reduction in incidence of NEC after probiotics

Before-after cohort study in a single NICUL acidophilus + B infantis: 250 million each daily

Int J Infect Dis 1999; 3:197-202

1993-94 No Probiotics

1994-95Probiotics

P value

n 1282 1237

NEC 85 (6.6%) 37 (3.0%) <0.0002

NEC-related Death

35 (2.7%) 14 (2.2%) < 0.005

Annual hospital charges associated with NEC in the US$290 – 772 million assuming

• 2,900 - 4,463 cases of NEC in US annually (1, 2) • $100,000 per case (1, 3) • $173,000 per case (4)

(1)Russell RB et al, Pediatrics 2007; 120: e1(2)Holman RC et al . Ped Perinat Epidemiol. 2006; 20: 498-506.(3)Christensen RD et al. Fetal Pediatr Pathol. 2010; 29: 185-98.(4)Bisquera JA, et al. Pediatrics. 2002; 109: 423-8.

J A Stockman IIIEditor: Yearbook of Pediatrics

‘One could not fault individual centres from proceeding with probiotics use until more data becomes available’

Yearbook of Pediatrics 2009;441-443

3. Better estimates of efficacy - ?

Up to 35% of controls in RCTs may become colonised by probiotics.The true effect of probiotics in reducing mortality may thus be underestimated by placebo RCTs.Kitajima H, et al. Arch Dis Child Fetal Neonatal. 1997; 76:F101-7.

Costeloe KL, et al. Pediatr Res. 2004; 55:2802 (Pt 2802 Supp S).

3. Better estimates of efficacy - ?

‘Before-after’ cohort studies may enhance the validity of placebo RCTs by providing uncontaminated estimates of efficacy, without colonisation of prior controls

Black N. Why we need observational studies to evaluate the effectiveness of health care. BMJ. 1996; 312:1215-8.

4. Pooling RCTs of different products is unacceptable - ?

Pooling is accepted in Cochrane Reviews of antibiotics, β blockers, tocolytics, calcium channel blockers, contraceptives, surfactants, hypothermia, ventilation, immunoglobulins and many others.

5. Better estimates of risk - ? • Potential risks of probiotics, such as sepsis,

allergy, antibiotic resistance are important.

• But these must be balanced against current evidence showing a substantial increase in mortality in controls.

• Large cohort/ Phase IV studies may estimate risks more reliably than RCTs

6. The current meta analysis suffers from heterogeneity - X

• There is no evidence of heterogeneity.

• Relative risks for mortality in all RCTs are < 1.0 and confidence intervals overlap.

• I2 = 0.

7. Access to probiotics may help clinicians

gain experience and confidence - å But only half the infants will get

probiotics in RCTs.

• Special access schemes or non randomised prospective cohort or case-control studies could allow the option of access for all eligible infants.

8. Many clinicians remain substantially

uncertain about the evidence - √

• If so, a RCT is an appropriate option for those clinicians.

• However, clinicians may also wish –to ensure access to probiotics for

eligible infants –discuss the options with parents

and let them decide.

What should parents be told?

Parents cannot exercise informed consent and choice - whether in RCTs or cohort studies of routinely offered probiotics - without transparent information.

Information should be given in as much detail as for a placebo RCT – no double standard

Probiotic use would ideally be within a prospectively planned cohort study or RCT in a local or national or international network of NICUs

Key points for parents

“In randomized studies in over 2000 preterm babies, those given probiotics were half as likely to die or to get necrotizing enterocolitis (a severe gut disease).”

Key points for parents

“There was no increase in adverse effects, but these are still possible, particularly in the smallest infants.”

Can we reach a consensus for global

collaboration?

Should placebo RCTs continue? YES

– If clinicians are unconvinced by the evidence, or

– If the only access to a previously evaluated probiotic product, shown to be effective in an earlier RCT, is through a placebo RCT

and

– If parents are fully informed and give consent

Can a previously evaluated probiotic be made available by a special access scheme?

YES - in certain jurisdictions, with

• Drug and Therapeutics Committee or IRB/ Executive approval• A reputable supplier of a previously evaluated probiotic

produced according to stringent guidelines for Good Manufacturing Practice

• Quality control by an accredited food standards laboratory with- random checks of each batch for extraneous pathogens- confirmation of colonisation in stools

• Detailed information to parents and their informed consent

Options for clinicians and parentsThose not convinced may wish to support ongoing placebo RCTs

Those who are convinced may wish to use a previously evaluated probiotic (e.g. Bin Nun et al, 2005 or Lin et al, 2008) in previously published doses, after informed parental consent, and preferably in a prospectively planned cohort study within a local or national network or registry

Danish Cohort Study of probiotic prophylaxis

• Inception date: March 2010• Infants < 30 weeks gestation in centres in

Denmark• Intervention: L acidophilus, BB 12 (produced by

Chr. Hansen company under EC Guidelines for Good Manufacturing Practice) offered to parents of eligible infants

• Primary outcomes: NEC, mortality

Other options for international RCTs

Multiple arm, ‘head to head’ RCTs with no placebo

Other options for international RCTs

Cluster factorial RCTs in which individual NICUs are randomly allocated to use , for example, one of various probiotic regimens, started early vs late

A workshop on global clinical collaborative probiotic research?Appropriate – as probiotics may be the most significant clinical advance in neonatology since surfactant

Important to seek representation from– Parent liaison groups– Drug regulatory and food standard agencies– International collaborative NICU networks– Bio-ethicists– Bio-statisticians, trialists, epidemiologists

Sir Iain ChalmersEditor, James Lind Library

Current attitudes to, and restrictions on, therapeutic research are powerful disincentives to people who wish to confront uncertainties about the effects of treatments.It is up to clinicians, patients, and the public in general to decide whether they wish to continue tolerating this bizarre state of affairs.

Chalmers I. BMJ 2008; 337: a841

Appendices

Rationale for further placebo RCTs

Issue Response CommentIdentify the optimum regimen

? • 2 arm placebo RCTs cannot reliably identify the optimum regimen. • This requires multiple arm RCTs

Confirm↓ mortality ? New RCTs are very unlikely to nullify

current evidence of lower mortality

Better estimates of efficacy

? • Placebo RCTs may underestimate efficacy if there is contamination of controls. • ‘Before-after’ cohort studies may add independent validity through uncontaminated estimates of efficacy.

Issue Response CommentPooling RCTs of different probiotics is unacceptable

? Pooling is accepted in Cochrane Reviews of antibiotics, β blockers, tocolytics, surfactants, ventilation immunoglobulins and many others.

Better estimates of short or long term risk

? • Potential risks of probiotics must be balanced against the substantially increased risk of mortality in controls.• Large cohort/ Phase IV studies may estimate risks more reliably than RCTs

Rationale for further placebo RCTs

Issue Response CommentCurrent meta analysis shows heterogeneity

x There is no evidence of heterogeneity. Relative risks for mortality are < 1.0 and confidence intervals overlap. I2 = 0.

Access to probiotics may help clinicians gain experience

√ • But only half the infants get probiotics in RCTs.• Special access schemes + cohort studies allow access for all infants.

Clinicians remain substantially uncertain about the evidence

√ • Should clinicians discuss the options with parents and let them decide?

Rationale for further placebo RCTs

Ioannidis and Lau, PNAS 2001

Analysed cumulative meta analyses of 45 different treatments in perinatal medicine.

Looked at how often there were big swings in the odds ratio for a treatment after a new trial was added to the analysis

“With 500 randomized patients, odds ratios in the range of 0.6–1.7 can easily be dissipated by future evidence – big swings are quite common.”

“With > 2000 patients ... odds ratios (or Relative Risks) may still change by as much as 0.74- to 1.35-fold, but bigger swings have not been seen.”

Current evidence of lower mortality is very unlikely to be nullified by more placebo RCTs.

Cumulative meta-analyses of Magnesium Sulphate in Acute Myocardial Infarction and albumin in adult intensive care were shifted to the null by RCTs of 58,050 and 6,997. Neither was associated with a swing in odds ratio of > 2.0.

Reduction in NEC in VLBW infants in a before-after cohort study in Japan

Controls (1994 - 98)

Probiotic (1999 - 03)

P value

n 226 338NEC 6 (2.6%) 0 (0%) <0.01Death 38 (16.8%) 39 (11.5%) NS

Satoh Y et al Int J Pro and Prebiotics 2007

Increased NEC in VLBW infants associated with probiotic in Finland over 12 years

• Incidence was highest in NICU using prophylactic Lactobacillus Rhamnosus GG 6 billion per day

• No before-after comparison

Luoto R, et al. Acta Paediatrica 2010; 99:1135–1138

Prophylactic LGG

Probiotic ‘on demand’

No probiotics

P value

19 ⁄ 418 (4.6%)

18 ⁄ 1024 (1.8%)

61 ⁄ 1900 (3.2%)

0.0090