Immunostimulants for preventing respiratory infections in children

EVIDENCE-BASED CHILD HEALTH: A COCHRANE REVIEW JOURNAL
Evid.-Based Child Health 7:2: 629–717 (2012)
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ebch.1833

Immunostimulants for preventing respiratory tract infection
in children (Review)

Del-Rio-Navarro BE, Espinosa-Rosales FJ, Flenady V, Sienra-Monge JJL

This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library
2011, Issue 6
http://www.thecochranelibrary.com

Immunostimulants for preventing respiratory tract infection in children (Review)
Copyright © 2011 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.


TABLE OF CONTENTS
HEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
PLAIN LANGUAGE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 632
SUMMARY OF FINDINGS FOR THE MAIN COMPARISON . . . . . . . . . . . . . . . . . . . 632
BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635
OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637
METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 637
Figure 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639
Figure 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 640
Figure 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 641
Figure 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 642
RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 642
Figure 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 646
Figure 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 647
Figure 7. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 648
Figure 8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 648
DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 649
AUTHORS’ CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652
CHARACTERISTICS OF STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 661
DATA AND ANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 693
Analysis 1.1. Comparison 1 Any IS compared with placebo, Outcome 1 Mean number of ARTIs. . . . . . . . 695
Analysis 1.2. Comparison 1 Any IS compared with placebo, Outcome 2 Percent difference in ARTIs. . . . . . . 697
Analysis 2.1. Comparison 2 Bacterial IS compared with placebo, Outcome 1 Mean number of ARTIs. . . . . . 699
Analysis 2.2. Comparison 2 Bacterial IS compared with placebo, Outcome 2 Percent difference in ARTIs. . . . . 700
Analysis 3.1. Comparison 3 Bacterial IS trials with n equal to or greater than 40 compared with placebo, Outcome 1 Mean
number of ARTIs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 701
Analysis 3.2. Comparison 3 Bacterial IS trials with n equal to or greater than 40 compared with placebo, Outcome 2
Percent difference in ARTIs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 702
Analysis 4.1. Comparison 4 Bacterial IS trials with n equal to or greater than 40 only OM-85 and BV D53 compared with
placebo, Outcome 1 Mean number of ARTIs. . . . . . . . . . . . . . . . . . . . . . . . 704
Analysis 4.2. Comparison 4 Bacterial IS trials with n equal to or greater than 40 only OM-85 and BV D53 compared with
placebo, Outcome 2 Percent difference in ARTIs. . . . . . . . . . . . . . . . . . . . . . 705
Analysis 5.1. Comparison 5 Adverse events, Outcome 1 Gastrointestinal adverse events. . . . . . . . . . . 706
Analysis 5.2. Comparison 5 Adverse events, Outcome 2 Skin adverse events. . . . . . . . . . . . . . . 707
Analysis 6.1. Comparison 6 OM-85 trials, Outcome 1 Mean number of ARTIs. . . . . . . . . . . . . . 708
Analysis 6.2. Comparison 6 OM-85 trials, Outcome 2 Percent difference in ARTIs. . . . . . . . . . . . . 709
Analysis 7.1. Comparison 7 D53 trials, Outcome 1 Mean number of ARTIs. . . . . . . . . . . . . . . 710
Analysis 7.2. Comparison 7 D53 trials, Outcome 2 Percent difference in ARTIs. . . . . . . . . . . . . . 711
ADDITIONAL TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711
APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713
WHAT’S NEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
HISTORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
CONTRIBUTIONS OF AUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
DECLARATIONS OF INTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 716
SOURCES OF SUPPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 717
INDEX TERMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 717

Immunostimulants for preventing respiratory tract infection in children (Review) 630


[Intervention Review]

Immunostimulants for preventing respiratory tract infection
in children

Blanca Estela Del-Rio-Navarro1 , Francisco J Espinosa-Rosales2 , Vicki Flenady3 , Juan JL Sienra-Monge1

1 Departmentof Allergy and Immunology, Hospital Infantil de México “Federico Gómez”, Mexico City, Mexico. 2 Department of
Immunology, Instituto Nacional de Pediatría (National Institute of Pediatrics), México D.F., Mexico. 3 Translating Research Into Practice
(TRIP) Centre - Mater Medical Research Institute, Mater Health Services, Woolloongabba, Australia

Contact address: Blanca Estela Del-Rio-Navarro, Department of Allergy and Immunology, Hospital Infantil de México “Fed-
erico Gómez”, Dr. Marquez 162, Colonia de los Doctores, Mexico City, DF, CP 06720, Mexico. blancadelrionavarro@gmail.com.
blancadelrionavarro@yahoo.com.mx.

Editorial group: Cochrane Acute Respiratory Infections Group.
Publication status and date: New search for studies and content updated (no change to conclusions), published in Issue 6, 2011.
Review content assessed as up-to-date: 3 March 2011.

Citation: Del-Rio-Navarro BE, Espinosa-Rosales FJ, Flenady V, Sienra-Monge JJL. Immunostimulants for preventing res-
piratory tract infection in children. Cochrane Database of Systematic Reviews 2006, Issue 4. Art. No.: CD004974. DOI:
10.1002/14651858.CD004974.pub2.


ABSTRACT

Background

Acute respiratory tract infections (ARTIs) are a major cause of childhood morbidity and mortality. Immunostimulants (IS) may reduce
the incidence of ARTIs.

Objectives

To determine the efficacy and safety of IS in preventing ARTIs in children.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) 2011, issue 1, which contains the Acute Respiratory
Infections Group’s Specialised Register, MEDLINE (1966 to February week 4, 2011), EMBASE (1990 to February 2011), Google
Scholar (2009 to February 2011), Scopus (2009 to February 2011), PASCAL (1990 to February 2010), SciSearch (1990 to February
2010) and IPA (1990 to February 2010).

Selection criteria

We included all comparative randomized controlled trials (RCTs) which enrolled participants less than 18 years of age. The intervention
was IS medication, administered by any method, compared to placebo to prevent ARTIs.

Data collection and analysis

We analyzed the outcome on ARTIs both as the mean number of ARTIs by group and as a percent change in the rate of ARTIs. We
undertook meta-analyses using a random-effects model and presented results as mean differences (MD) with 95% confidence intervals
(CI). Two review authors independently assessed the search results and risk of bias, and extracted data. A funnel plot suggested there
may be publication bias in the identified trials.


Main results
Thirty-five placebo-controlled trials (4060 participants) provided data in a form suitable for inclusion in the meta-analyses. When
compared with placebo, the use of IS was shown to reduce ARTIs measured as the total numbers of ARTIs (MD -1.24; 95% CI -1.54
to -0.94) and the difference in ARTI rates (MD -38.84%; 95% CI -46.37% to -31.31%). Trial quality was generally poor and a high
level of statistical heterogeneity was evident. The subgroup analysis of bacterial IS, D53 and OM-85 studies produced similar results,
with lower heterogeneity. No difference in adverse events was evident between the placebo and IS groups.
Authors’ conclusions
This review shows that IS reduce the incidence of ARTIs by 40% on average in susceptible children. Studies in healthy children are not
available. Although the safety profile in the studies was good, some IS may be unsafe. ARTI-susceptible children may benefit from IS
treatment. Further high-quality trials are needed and we encourage national health authorities to conduct large, multicentre, double-
blind, placebo-controlled RCTs on the role of IS in preventing ARTIs in children.

PLAIN LANGUAGE SUMMARY
Immunostimulants to prevent acute respiratory tract infections in children
Acute respiratory tract infections (ARTIs) are responsible for 19% of all deaths in children younger than five years of age, mainly in low-
income countries in Africa, Asia and Latin America. In high-income countries ARTIs are among the most frequent illnesses, leading
to 20% of medical consultations, 30% of days lost from work and 75% of antibiotic prescriptions. In the USA the total cost of non-
influenza-related viral ARTIs is around $40 billion annually, while the corresponding cost for influenza is US $87.1 billion. The main
signs and symptoms of ARTIs include sneezing, runny nose, sore throat, cough and malaise. Children living in rural communities, not
attending daycare centres, suffer about seven ARTI episodes in the first year of life; eight ARTIs per year from the ages of one to four;
six per year aged five to nine; and five per year aged 10 to 19. Children exposed to risks factors, such as attendance at daycare centres,
overcrowding, contact with older siblings, smoking at home and lack of breast feeding, may suffer more ARTIs.
Several treatments have been used to reduce the incidence of ARTIs (vitamin A, vitamin C, zinc, antibiotics). Among them are
immunostimulants (herbal extracts, bacterial extracts, synthetic compounds), which aim to increase the immune defences of the
respiratory tract. We searched for clinical trials of immunostimulants to prevent ARTIs in children compared to placebo. Our review
includes 35 studies with 4060 participants. However, the quality of many of the studies was poor and the results were very diverse.
By combining results, immunostimulants reduced 1.24 ARTIs in a six-month period, equivalent to a 39% reduction in ARTIs compared
to the placebo group. Only 20 studies provided adequate data on adverse events: the most frequent were rash, nausea, vomiting,
abdominal pain and diarrhea. The main limitations of this review were the poor methodological quality and diverse trial results.
We conclude that ARTI-susceptible children may benefit from immunostimulants, but more high-quality studies are needed. We
suggest that national health authorities conduct high-quality randomized controlled trials to assess the true effects of immunostimulant
preparations.



S U M M A R Y O F F I N D I N G S F O R T H E M A I N C O M P A R I S O N [Explanation]

Any immunostimulant (IS) compared with placebo for preventing respiratory tract infection in children

Patient or population: children (age <18 years) susceptible to acute respiratory tract infections (ARTIs)
Settings: outpatient
Intervention: any IS
Comparison: placebo

Outcomes Illustrative comparative risks’ (95% CI) No of participants (studies) Quality of the evidence Comments
(GRADE)

Assumed risk Corresponding risk

Placebo Any IS

Number of ARTIs The range of ARTIs in the con- The mean Number of ARTIs in 4060 ⊕⊕⊕ The effect depends on the
trol group was 0.92 to 6.2 the intervention groups was (35 studies) moderate1 number of ARTIs in the control
1.24 lower (0.94 to 1.54 group
lower)

Percent difference in ARTIs The mean Percent difference 4060 ⊕⊕⊕
in ARTIs in the intervention (35 studies) moderate1,2
groups was
39 lower (31.31 to 46.37
lower)

Incidence of gastrointestinal 21 per 1000 30 per 1000 (11 to 50 per 1457 ⊕⊕
adverse events 1000) (10 studies) low1,3

Incidence of skin adverse 3 per 1000 7 per 1000 (-8 to 14 per 1000) 1469 ⊕⊕
events (10 studies) low1,3

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the
assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; RR: risk ratio; OR: odds ratio
633


GRADE Working Group grades of evidence
High quality: further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: we are very uncertain about the estimate.

1 Funnel plot shows possible publication bias, risk of bias in the studies moderate, and high heterogeneity among studies. A group of six studies with good quality point to the benefit of IS
2 Heterogenity decreased with calculation of percent difference ARTIs.
3 Adverse events were reported only in 10 trials; selective outcome reporting

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
634

BACKGROUND
effort to establish the aetiology and epidemiology of ARTIs in
children in high-income countries. The project was carried out
Description of the condition in populations from 0 to 59 months of age in Africa, Asia and
Latin America. The incidence rate in six community-based stud-
In 1998 the World Health Organization (WHO) considered acute
ies ranged from 12.7 to 16.8 ARTIs per 100 child-weeks and the
respiratory tract infections (ARTIs) to be “the forgotten pandemic”
incidence of lower ARTIs was from 0.2 to 0.4 per 100 child-
as ARTIs caused 19% of all deaths in children younger than five
weeks. The children studied spent from 21.7% to 40.1% of the
years and 8.2% of all disabilities and premature mortality (WHO
observed weeks with ARTIs and from 1% to 14.4% of the ob-
1998). In 2000, 1.9 million (95% confidence interval (CI) 1.6 to
served weeks with lower ARTIs. Viral agents were more frequently
2.2 million) children died from ARTIs worldwide, with 70% of
recovered than bacterial agents; respiratory syncytial virus (RSV)
the deaths in Africa and South-East Asia (Williams 2002). AR-
was the most frequent virus (Selwyn 1990). In Mexico, a study
TIs are the leading cause of morbidity in high-income countries
assessing the effect of daycare centres on ARTI incidence followed
(USA, Canada, Western Europe) and account for 20% of medical
144 children (aged 43 days to 4 months at entry) at home for one
consultations, 30% of days lost from work and 75% of antibi-
year. The study found that these children had six ARTIs each year,
otic prescriptions (WHO 1998). ARTIs are responsible for most
with a median of 40 sick days in a year (Flores-Hernandez1999).
sick days amongst school children (Haskins 1986) and parental
Viruses were the main aetiological agents for ARTIs in children
absenteeism from work (Bell 1989). The cost of ARTIs in high-
at daycare centres (Denny 1986) and in the community (Monto
income countries is significant. For example, between 2000 and
1993). The most common virus isolates are rhinovirus, respiratory
2002 there were approximately 500 million non-influenza-related
syncytial virus, parainfluenza virus and adenovirus. Lower ARTIs
viral ARTI episodes in the USA per year; the total economic im-
are also frequently associated with viral infections, but bacterial
pact of these episodes was around $40 billion annually (direct costs
agents may be found in 4.5% to 40% of the cases (Selwyn 1990).
of $17 billion per year and indirect costs of $22.5 billion per year)
Up to 50% of children admitted to hospital with proven bacte-
(Fendrick 2003). Based on the 2003 US population, it has been
rial ARTIs also have evidence of concurrent or recent viral ARTIs
calculated that the total economic burden of annual influenza epi-
(Campbell 1995). The damage caused by viruses to epithelial cells
demics in the USA was $87.1 billion US dollars ($47.2 to $149.5)
in the airways may increase the adherence of bacteria and lead to
(Molinari 2007). Risk factors for ARTIs in childhood include at-
a bacterial superinfection (Hament 1999).
tendance at daycare centres (Schwartz 1994), overcrowding (Bell
History of respiratory infections in the first 12 years among chil-
1989; Selwyn 1990), contact with elder siblings (Selwyn 1990),
dren was established in a cohort of German children living in ur-
male gender (Monto 2002), smoking at home (Jin 1993) and lack
ban areas. The mean cumulative number of ARTIs in the 12 years
of breast feeding (Wright 1989).
was 21.9 (standard deviation (SD) 9.0) episodes; the mean annual
Community health studies in high-income countries have pro-
number was 1st year, 3.1 (2.1) episodes; 2nd year, 3.2 (2.5); 3rd
vided basic information on the incidence of ARTIs in children.
year, 2.1 (2.0); 4th year, 2.3 (2.1); 5th year, 1.8 (1.6); 7th through
In the Cleveland Family Study, 100 families were visited weekly
to the 9th year, 1.1 (1.0) episodes; 10th year through 12th year,
by nurses during 1948 to 1957. The annual frequency of respi-
1.0 (0.9) episodes (Grüber 2008). The frequency of ARTIs in this
ratory illness was 6.72 in children less than one year old; 7.95 in
study was about a half of the classic Monto studies (Monto 1974;
children aged one to four years; 6.21 in children aged five to nine
Monto 1993). The authors regarded incidence above the twofold
years; 5.02 in children aged 10 to 14 years; and 4.71 in 15 to 19-
standard deviation as clinically relevant; more than seven episodes
year olds (Monto 2002). In the first phase of the Tecumseh study,
in the 1st year of life, more than eight episodes in year two, more
from 1965 to 1971, 4905 residents registered their incidence of
than six episodes in year three and year four more than five episodes
ARTIs for the six-year period. In the first report the residents had
in year five, more than four episodes in year six, and more than
suffered approximately 14,600 ARTIs. The annual incidence of
three episodes from year seven onwards.
ARTIs per person was 6.1 in children less than one year old; 5.7 in
In a healthy population without any special risk factors or immun-
children aged one to two years; 4.7 in children aged three to four
odeficiencies, there is a subgroup of people with a higher incidence
years; 3.5 in children aged five to nine years; 2.7 in children aged
of ARTIs. A cohort of children from Nijmegen, Netherlands was
10 to 14 years old; and 2.4 in 15 to 19-year olds (Monto 1974).
followed for 21 years to register the occurrence of ARTIs. The
In the second report of the Tecumseh study, which covered two
number of respiratory infections was assessed at the ages of two,
phases comprising a total of 11 years (1965 to 1971 and 1976 to
four, eight and 21. It was considered that a person had a recurrent
1981), the mean annual number of ARTIs was 4.9 in the group
infection if the number of ARTIs was above 75th percentile of the
aged from zero to four years; and 2.8 in the group aged five to 19
distribution of respiratory infection at each assessment. Twenty-
years (Monto 1993). In both phases of the study viruses were the
three percent of the people had recurrent respiratory infection in
most common agents causing ARTIs.
two or more assessments and 1% suffered from recurrent infection
During the 1980s, the Board on Science and Technology for In-
in the four assessments (Rovers 2006).
ternational Development (BOSTID) undertook a co-ordinated



Recent influenza epidemics have increased interest in effective un- The actual mechanism of IS is not yet fully understood. Currently
specific measures to protect the global population, outside the pro- the mechanisms of action are known for only two synthetic IS,
duction of appropriate vaccines which could take more than six tucaresol and imiquimod. The mechanism of tucaresol is to form a
to nine months to be ready for use. These unspecific measures in- complex on the surface of T cells (a Schiff reaction with the amines,
clude physical methods to reduce the spread of respiratory viruses probably on CD2). This complex provides an additional stimu-
such as hand washing, wearing masks, gloves and gowns (Jefferson lant which facilitates the activation of the T cells (a co-stimula-
2009). In military populations, reported additional measures to tory signal activating the MAPK ERK2 pathway) (Rhodes 1995).
prevent respiratory tract infections include reducing contact be- Imiquimod and other related molecules activate the immune cells
tween units, reducing crowding, installing cloth barriers between by binding to the receptor for the bacterial products that activate
beds, indoor air dilution and ventilation, dust suppression and the unspecific defence mechanism and promote the immune re-
air sterilisation (Lee 2005). Other measures include vitamin and sponse; they bind to the Toll-like receptor 7 (TLR7) activating the
mineral supplementation, such as vitamin A (Chen 2008a), vita- MyD88-dependent signalling pathway (Hemmi 2002).
min C (Hemilä 2010), vitamin D (Yamshchikov 2009) and zinc It may be postulated that products with IS properties activate the
(Aggarwal 2007). Interventions which stimulate the immune sys- immune cells using the receptors that recognise common bacte-
tem (immunostimulants) have been proposed as effective measures rial products or receptors that provide additional stimulation for
to reduce ARTIs. activation. For instance, Toll-like receptors (TLR) recognise com-
Some years ago, the idea that bacterial lysates, plant extracts or ponents common to a range of bacteria, so-called pathogen-as-
imidazole compounds, could induce unspecific immunity against sociated molecular patterns (PAMPs) such as lipopolysaccharide,
viruses and distinct bacteria was not very solid. However, the re- peptidoglycan, lipoteichoic acid, lipoarabinomannan, un-methy-
cent discovery of Toll-like-receptors (TLRs) supports the possible lated DNA with CpG motif and bacterial lipoproteins which ac-
mechanism of action of immunostimulants (Krieg 2003). TLRs tivate the innate immune responses. The innate immune response
were discovered in the 1990s and their importance on immunity is responsible for the early mechanisms of defence against infec-
was found later, see How the intervention might work below. In tion; for instance the phagocytosis and neutralisation of bacte-
fact, there is evidence that two bacterial lysates may act on TLR2 ria entering the body. The mechanisms that enhance the innate
(Alyanakian 2006; Nikolova 2009), as well as levamisole (Chen immune responses (cytokines and chemokines) also stimulate the
2008b). adaptive immune response (production of specific antibodies and
reproduction of specific T cells) (Hoffmann 1999; Schnare 2001;
Takeuchi 2001). In fact, there is evidence that two bacterial lysates
Description of the intervention may act on TLR2 (Alyanakian 2006; Nikolova 2009), as well as
levamisole (Chen 2008b).
The main way to prevent ARTI complications is to prevent these
infections and administer early antibiotic treatment when bac-
terial ARTIs are diagnosed (Heikkinen 1999; Henderson 1982;
WHO 1998). Non-specific preventative measures for ARTIs stud-
Why it is important to do this review
ied in clinical trials include general hygiene methods in children
attending daycare centres (CDCIDSG 1984); the administration Most ARTIs are caused by viruses, hundreds of which may cause
of nutritional supplements such as vitamin A to malnourished this type of infection. It would be impractical, therefore, to have
children (Barreto 1994), vitamin C to normal and malnourished a vaccine for each possible pathogenic agent. Therefore, spe-
children (Hemila 1997; Jefferson 2001) and trace elements to cific immunisation may not be the ultimate solution to prevent
malnourished and susceptible children (Sazawal 1998); preven- ARTIs. The introduction of the pneumococcal conjugate vac-
tive antibiotics (Dajani 1995); administration of gamma glob- cine decreased carriage and invasive infections due to the vaccine
ulins (Nydahl-Persson 1995); nasal spray of immunoglobulins serotypes, but it has been replaced by other non-vaccine serotypes
(Heikkinen 1998); herbal extracts (Grimm 1999); xylitol sugar that are becoming antibiotic resistance (Hsu 2009; Huang 2009;
syrup or chewing gum (Uhari 1998); and the use of immunos- Mera 2009).
timulants (IS) from different sources. The sources are synthetic IS could provide an alternative to vaccines for preventing AR-
(Passali 1994a); thymic extracts or factors (De Mattia 1993); or of TIs but the efficacy of these medications is controversial (Collet
biological origin such as Klebsiella extracts containing lipopolysac- 1992; Valleron 1992). Several bacterial extracts and synthetic com-
charide (Dahan 1986) and mixtures of bacterial extracts (Berber pounds are used in Europe and the Americas to prevent ARTIs.
1996). However, the evidence of the safety and efficacy of this approach
is unclear. A systematic review of immunostimulation for the pre-
vention of ARTIs in children is required to enable a robust ap-
How the intervention might work praisal of the current evidence on the safety and efficacy of this
approach and to provide clues for the development of new IS.



OBJECTIVES Search methods for identification of studies
To assess the safety and efficacy of immunostimulants (IS) admin-
istered to children to prevent ARTIs when compared with placebo,
in terms of frequency of these infections and reported adverse ef- Electronic searches
fects. Trials comparing two IS were also included.
For this update we searched the Cochrane Central Register of Con-
trolled Trials (CENTRAL) 2011, issue 1, which contains the Acute
Respiratory Infections Group’s Specialised Register, MEDLINE
METHODS (1966 to February week 4, 2011), EMBASE (1990 to February
2011), Google Scholar (2009 to February 2011), Scopus (2009
to February 2011), PASCAL (1990 to February 2010), SciSearch
(1990 to February 2010) and IPA (1990 to February 2010). De-
Criteria for considering studies for this review
tails of the previous searches are in Appendix 1.
We used the following search strategy to search MEDLINE and
CENTRAL. To search MEDLINE, we combined the search strat-
Types of studies egy with the Cochrane Highly Sensitive Search Strategy for identi-
Randomised controlled trials (RCTs) comparing IS, administered fying randomized trials in MEDLINE: sensitivity- and precision-
by any method, to placebo to prevent ARTIs. Trials referring to maximising version (2008 revision); Ovid format (Lefebvre 2009).
interferon inducers, vitamins and nutritional supplements were The search strategy was adapted for EMBASE (see Appendix 2).
not included. Details of the PASCAL, SciSearch, IPA and previous Embase
search are in Appendix 3.
MEDLINE (Ovid)
Types of participants 1 exp Respiratory Tract Infections/
2 (respiratory adj5 infection*).tw.
Participants younger than 18 years of age. We did not include trials 3 1 or 2
that included participants who suffered from asthma, allergy and 4 exp Adjuvants, Immunologic/
atopy, or chronic respiratory diseases. 5 immunostimulant*.tw,nm.
6 immunomodulat*.tw,nm.
7 immunoadjuvant*.tw,nm.
Types of interventions
8 immunologic adjuvant*.tw,nm.
The use of an IS administered by any method to prevent ARTIs. 9 (immunobalt or lw50020 or luivac or paspat or munos-
Administration of IS could begin in the presence of active ARTI. tin).tw,nm.
We considered trials utilising concomitant therapies such as an- 10 (om-85 bv or om85bv or om 85 bv).tw,nm.
tipyretics or antibiotics for inclusion. 11 (bronchovaxom or broncho-vaxom or broncho vaxom).tw,nm.
12 (pulmonar-om or pulmonar om).tw,nm.
13 d53.tw,nm.
Types of outcome measures 14 (ribomunyl or ribovac or immucytal).tw,nm.
A broad definition of ARTI was accepted and included using differ- 15 Lipopolysaccharides/
ent specific diagnoses, such as cold, influenza, tonsillitis, pharyngi- 16 lipopolysaccharide*.tw,nm.
tis, bronchitis and otitis media. Aetiological agents were not stud- 17 (ru41740 or ru-41740 or ru 41740 or biostim).tw,nm.
ied and no distinction was made between bacterial and viral AR- 18 Thymus Extracts/
TIs. Physician diagnosis of ARTI and adverse events was accepted. 19 thymus extract*.tw,nm.
20 (thymic extract* or thymomodulin*).tw,nm.
21 Pelargonium/
Primary outcomes 22 (pelargonium* or umckaloabo).tw,nm.
23 (am3 or imunoferon or immunoferon or inmunoferon).tw,nm.
The number of ARTIs in children suffered during the study period. 24 glycophosphopep*.tw,nm.
25 (pidotimod or adimod).tw,nm.
26 Levamisole/
Secondary outcomes
27 levamisole.tw,nm.
1. The percentage of ARTIs. 28 or/4-27
2. The incidence of adverse events. 29 3 and 28



Searching other resources Saracho-Weber 2001 (co-worker Vázquez-Ramos); Schaad 1986;
We used identiﬁed articles as references for a Science Citation Schaad 2002). However, no additional data were provided. A fur-
Index search. We searched bibliographies of all included trials as ther 11 trial authors were contacted by Dr Berber without response
well as those of relevant reviews to identify additional studies. (Aymard 1994; Careddu 1994a; Careddu 1994b; Fiocchi 1986;
Finally, we sent a letter to all ﬁrst authors, as well as pharma- Fiocchi 1988; Fiocchi 1989; Fiocchi 1990; Motta 1994; Paupe
ceutical companies that manufacture immunostimulant drugs, re- 1986; Rutishauser 1998 (co-worker Grevers); Valleron 1992). Dr.
questing data and references for any relevant published and un- Arturo Berber provided the database for OM-85 trials from Mex-
published trials. There were no language or publication restric- ico. In 2010, we made attempts to contact the following au-
tions. We also searched for studies in the trial registration web thors: Joseph Bellanti, Jean Bousquet, Herman A. Cohen, Craig
site: metaRegister of Controlled Trials (http://www.controlled- I Coleman, Jean Paul Collet, Alessandro Fiocchi, Sergio Mar-
trials.com/mrct/). We searched for IS trial registries in the U.S. cassa, Renzo Mora, RJ Riedl-Seifert, Urs B. Schaad, Draganka
National Institutes of Health in http://www.ClinicalTrials.gov. Stankulova, Claudia Steurer-Stey and James A. Taylor, and manu-
facturers Luipold (luivac), OM Pharma (broncho-vaxom), Pierre
Fabre (ribomunyl) and Polichem (adimod). Only Sergio Marcassa,
Renzo Mora, RJ Riedl-Seifert, Urs B. Schaad (by himself and in
Data collection and analysis
name of OM Pharma) replied; no information regarding new stud-
ies was obtained.
Selection of studies
Two review authors (BN, JSM) independently searched for trials Assessment of risk of bias in included studies
for inclusion and risk of bias assessment. We resolved differences
by discussion. We measured trial quality using seven domains.
1. Random sequence generation (selection bias).
2. Allocation concealment (selection bias).
Data extraction and management 3. Blinding (performance bias and detection bias).
We analyzed and managed data using Review Manager (RevMan 4. Blinding of participants and personnel (performance bias).
2008). Two authors (BN, JSM) independently extracted data. We 5. Blinding of outcome assessment (detection bias).
sought missing data from investigators of individual trials, as nec- 6. Incomplete outcome data (attrition bias).
essary, in order to perform analyses on an intention-to-treat (ITT) 7. Selective reporting (reporting bias).
basis. We assigned a quality rating for above domains for each included
Dr Arturo Berber contacted trial authors to request unpublished trial using the criteria outlined in the Cochrane Handbook for Sys-
data. Responses were received from 10 trial authors (Arroyave tematic Reviews of Interventions (Higgins 2011) as high risk, low
1999; Collet 1993; Gómez-Barreto 1998; Gutiérrez-Tarango risk or uncertain risk . Figure 1 and Figure 2 shows the results of
2001; Jara-Pérez 2000; Karam-Bechara 1995; Paupe 1991; risk of bias assessment for the seven domains.



Figure 1. Risk of bias graph: review authors’ judgements about each risk of bias item presented as
percentages across all included studies.



Figure 2. Risk of bias summary: review authors’ judgements about each risk of bias item for each included
study.



The way in which the outcomes were reported varied widely across
Measures of treatment effect
the trials. We decided to use the mean number of ARTIs and its
We reported the mean differences (MD) (and 95% conﬁdence standard deviation (SD) as the outcome as it allows the use of para-
intervals (CI)) for the meta-analysis of data measured on a contin- metric statistical methods that provide more power to the tests. We
uous scale. We assessed heterogeneity by visual inspection of the assumed that the number of ARTIs in the IS-treated group would
outcomes tables and by using two statistics of heterogeneity (H be comparable to the number of ARTIs in the placebo group; and
and I2 statistic) (Higgins 2003). Due to the observed statistical both of these would depend on the susceptibility of the children
heterogeneity, we used the random-effects model. (determined by age, duration of trial and seasons of the year dur-
ing the trial). Consequently we expected to have heterogeneity in
the mean number of ARTIs. Therefore, we decided to standardise
Unit of analysis issues
the results using the percentage of infections considering the mean
Regarding the trials with a description of randomisation and al- number of infections in the placebo group as 100%.
location, the unit of randomisation was the individual subject
(Cohen 2004; Collet 1993; Del-Rio-Navarro 2003; Gutiérrez-
Tarango 2001; Jara-Perez 2000; Taylor 2003). Assessment of reporting biases
We assessed publication bias using a funnel plot (Egger 1997). The
results indicated possible publication bias. The funnel plot for the
Dealing with missing data
number of ARTIs was asymmetrical with a large base skewed to
The studies only analyse the available data, ignoring the missing the left and narrower distribution at the top; the funnel plot for
data. the percentage of ARTIs was more symmetrical but the most of
the points were on the left side. Figure 3 and Figure 4 are funnel
plots showing the differences in the number of ARTIs and the
Assessment of heterogeneity
percentage of ARTIs.

Figure 3.



Figure 4.

than 40 including only OM-85 and BV D53 and OM-85 alone
Data synthesis
and D53 alone.
Due to the heterogeneity of the results we selected the random-
effects model of meta-analysis.

Subgroup analysis and investigation of heterogeneity RESULTS
We also conducted bivariate correlation as well as linear regression
and sensitivity analyses of subgroups to investigate the sources of Description of studies
statistical heterogeneity (please refer to Results section).
See: Characteristics of included studies; Characteristics of excluded
studies.
Sensitivity analysis
To determine whether conclusions were robust, we performed
analyses of different set of studies as follows: any IS; bacterial IS; Results of the search
bacterial IS trials with the total sample size of equal to or greater The electronic search produced 764 references. No other poten-
than 40; bacterial IS trials with total sample size equal to or greater tially eligible studies were found as a result of contact with the trial


authors or searching of trial registries. Of the references obtained, the ARTIs were defined by the presence of respiratory signs and
we identified 93 studies as potentially eligible. symptoms.
The remaining 24 studies reported a variety of end points: symp-
toms, clinical scales or presence or absence of respiratory infec-
Included studies tions. Some trials reported the frequency of ARTIs as either equal
We included 61 placebo-controlled clinical trials involving 4149 to or more than one infection (Burgio 1994; Careddu 1994b;
participants. The studies were very heterogenous in the interven- Fukuda 1999; Mora 2010a; Paupe 1991; Rutishauser 1998; Taylor
tions studied, the number of ARTIs in the placebo groups and in 2003; Wahl 2008); equal to or more than two infections (Mora
the reporting of outcomes. 2007); equal to or more than three infections (Collet 1993); the
total number of ARTIs; mean number of ARTIs (Caramia 1994;
Chen 2004; Dils 1979; Fiocchi 1988; Longo 1988; Passali 1994;
Population
Pozzi 2004; Riedl-Seifert 1995; Sramek 1986); or reduction in
The participants enrolled in the included trials were children rang- the number of ARTIs (Andrianova 2003; Iuldashev 1988). Others
ing from six months to 18 years of age. The echinacea trials differed measured the severity of symptoms using clinical scales that were
in the selection criteria of participants as they used children with- not validated (Fiocchi 1989; Giovannini 2000; Mora 2002; Renzo
out a significant health problem and without a history of recurrent 2004) or as days suffering symptoms (Martin du Pan 1982).
ARTI. The remainder of the trials included a history of recurrent
ARTIs in the inclusion criteria. All the trials were conducted in
the Northern (boreal) hemisphere except Fukuda’s (Fukuda 1999). Excluded studies
Fall and winter seasons referred to the months from September We excluded 36 studies: 32 did not comply with the selection
to January. The year of the trial was not specified in most of the criteria; two compared several IS treatments without a placebo
studies. group; and two were duplicate reports of trials already included.
See Characteristics of excluded studies table.
Interventions
Forty studies used bacterial products, four studies used herbal ex-
Risk of bias in included studies
tracts (echinacea and garlic), 11 studies used synthetic compounds,
five studies used thymic extracts (thymomodulin) and one study The description of the methodology was not clear in most of
used a synthetic interferon. All trials used a placebo control. The the studies. Only 17.1% (6 out of 35) studies reported adequate
common names of the medications are in Table 1. randomisation and blinding (participants and treating physicians
Twenty-two studies had a duration shorter than six months, 33 were blinded) (Cohen 2004; Collet 1993; Del-Rio-Navarro 2003;
studies had a duration of six months and only six studies had a Gutiérrez-Tarango 2001; Jara-Perez 2000; Taylor 2003) (Figure
duration longer than six months. The duration of seven D53 trials 1). Using the quality rating criteria (Higgins 2011), the quality of
was less than six months and nine D53 trials had a duration of six the rest of the trials (28 out of 34 (82.4%)) was B; randomisation
months. In all D53 trials the description of the methodology was and follow through of participants was unclear. See Table 2 for a
not clear and different routes of administration were used (nasal description of the quality of the trials.
spray or by mouth). Ten OM-85 BV trials lasted six months; two Only 17.1% (6 out of 35) studies reported on the number of
trials had a duration of longer than six months. participants lost to follow up (Cohen 2004; Collet 1993; Del-Rio-
Navarro 2003; Gutiérrez-Tarango 2001; Jara-Perez 2000; Taylor
2003). Of these, five studies reported losses. While losses were
Outcomes minimal, 0.5% to 7% in three studies, two studies reported a
Only 35 of the 61 included studies reported the mean and SD loss of 18% and 24%. As additional data were not obtained from
of the incidence of respiratory infections or provided data to cal- the investigators on the outcomes of participants who were lost
culate these measure, allowing their inclusion in the meta-analy- to follow up an intention-to-treat (ITT) analysis could not be
sis (Ahrens 1984; Arroyave 1999; Careddu 1994a; Clerici 1988; undertaken. The numbers lost to follow up are explained in further
Cohen 2004; De Loore 1979; Del-Rio-Navarro 2003; Fiocchi detail.
1986; Garabedian 1990; Gutiérrez-Tarango 2001; Gómez-Barreto In the Cohen 2004 trial, 160 out of 215 in the IS group com-
1998; Hauguenauer 1987; Hüls 1995; Jara-Pérez 2000; Karam- pleted the trial (27 dropped out because the medication had an
Bechara 1995; Lacomme 1985; Litzman 1999; Maestroni 1984; unpleasant taste, 24 due to non-compliance which was not fully
Motta 1994; Paupe 1986; Pech 1987; Piquett 1986; RB10 1994; explained, four dropped out due to a lack of confidence in the
RB17 1988; RB21 1988; RB22 1990; RB24 1990; RB25 1990; treatment); and 168 out of 215 of the placebo group completed
Saracho-Weber 2001; Schaad 1986; Schaad 2002; Van Eygen the trial (22 dropped out because the placebo had an unpleasant
1976; Van Eygen 1979; Vautel 1993; Zagar 1988). In these studies taste, 21 due to non-compliance which was not fully explained,



and one dropped out due to a lack of confidence in the treatment). tion system network in France). Additionally, randomisation was
A total of 24% were lost to follow up. also stratified by study centre and blocked for every four children.
In the Del-Rio-Navarro 2003 trial, 20 out of 25 in the IS group In the studies by Del-Rio-Navarro 2003 and Gutiérrez-Tarango
completed the trial. Five children were lost to follow up. Twenty 2001 consecutive numbered study medication boxes, as well as a
out of 24 in the placebo group completed the trial. Two children closed opaque envelope describing the treatment, were supplied
were lost to follow up (the parents of one participant withdrew directly by the manufacturer; participants received a patient num-
consent for their child to continue in the trial and one left because ber coincident with treatment number when the selection criteria
the trial medication caused the child to have diarrhea). A total of were completed. In the Jara-Perez 2000 study consecutive num-
18% were lost to follow up. bered study medication boxes, as well as a closed opaque enve-
In the Collet 1993 trial, 199 out of 210 in the IS group and 196 lope describing the treatment, were supplied directly by the man-
out of 213 placebo group completed the trial. For both groups, ufacturer; participants received a patient number corresponding a
the 28 lost to follow up were related either to the parents moving treatment according to an alphabetical name list. In Taylor 2003,
to a different location or the mothers stopped working and no each study centre had a supply of study medication (active medi-
longer took their children to the daycare centres where the trials cation and placebo) in consecutively numbered bottles that were
were being held. A total of 7% were lost to follow up. identical in appearance, contents of each bottle were randomly
In the Taylor 2003 trial, 242 out of 263 in the IS group completed determined using a computer-generated randomisation list, and
the trial (six withdrew before the first ARTI, five changed their randomisation was stratified by site and in blocks of 10. Enrolled
minds about participating, one never received the study medica- children were assigned a unique study number corresponding to
tion, five withdrew during the first ARTI, three refused the study the numbers on the bottles of study medication. Figure 1 shows
medication, one was concerned about the effect on their immune the review authors’ judgement of the risk of bias related to allo-
system, for one the protocol was too complicated, six log books cation concealment presented as percentages across all included
were never received, four were lost to follow up); 244 out of 261 studies and Figure 2shows the risk for each included study.
in the placebo group completed the trial (three withdrew before
the first ARTI, two changed their minds about participating, one
was excluded for taking another medication, one withdrew during Blinding
the first ARTI, nine log books were never received and four were Six studies (Cohen 2004; Collet 1993; Del-Rio-Navarro 2003;
lost to follow up). A total of 7% were lost to follow up. Gutiérrez-Tarango 2001; Jara-Perez 2000; Taylor 2003) claimed
In the Jara-Pérez 2000 trial, 99 out of 100 in the IS group com- that IS and placebo treatment had identical appearance and that
pleted the trial. The case report from one child was lost. One hun- the taste of both were similar. Investigators and participants were
dred out of 100 in the placebo group completed the trial. A total not aware of the received treatment.
of 0.5% were lost to follow up.
In the Gutiérrez-Tarango 2001 trial, outcomes were reported for
all enrolled children. All the participants completed the trial. Incomplete outcome data
The outcome assessor was blinded to the treatment allocation in Only six studies (Cohen 2004; Collet 1993; Del-Rio-Navarro
8.8% (three out of 34) of the studies (Cohen 2004; Collet 1993; 2003; Gutiérrez-Tarango 2001; Jara-Perez 2000; Taylor 2003) re-
Taylor 2003). Only 14.7% (five out of 34) of studies (Cohen 2004; ported the number of patients lost to follow up. All used only the
Collet 1993; Del-Rio-Navarro 2003; Gutiérrez-Tarango 2001; available data for the analyses. No imputation for the incomplete
Taylor 2003) were considered of quality A (Higgins 2011). The data were intended.
rest of the studies were of poor quality. Cohen 2004 had a total of 24% lost to follow up; the frequencies
and causes of it were similar in active and placebo groups. In
the Del-Rio-Navarro 2003 trial 18% of participants were lost to
Allocation follow up; more children in the active group were lost (5/25 in
In the studies with a proper description of randomisation and active group versus 2/25 in the placebo group). In the Collet 1993
allocation (Cohen 2004; Collet 1993; Del-Rio-Navarro 2003; trial, 7% of participants were lost to follow up; in both groups the
Gutiérrez-Tarango 2001; Jara-Perez 2000; Taylor 2003), the im- rates and the reasons were similar. In the Taylor 2003 trial, 7% of
plementation of the random sequence of the treatments was re- participants were lost to follow up; in both groups the rates and
ported. In Cohen 2004 active medication and the placebo were the reasons were similar. In the Jara-Pérez 2000 trial only the data
supplied directly by the manufacturer and all randomisation lots of one participant on IS was lost. In the Gutiérrez-Tarango 2001
were stored in a sealed envelope at the pharmacy of the company, trial, all the participants completed the trial.
to be opened only in the event of an emergency. In Collet 1993
participants were allocated to IS or placebo according to a pro-
Selective reporting
gram for remote data entry (Minitel a national telecommunica-



The study protocols were not available. Sixty-one randomized, non-indexed, small, local journals).
placebo-controlled clinical trials were identified. Only 35 stud-
ies reported the mean and SD of the incidence of respiratory
infections or provided data to calculate these measure (Ahrens Effects of interventions
1984; Arroyave 1999; Careddu 1994a; Clerici 1988; Cohen
2004; De Loore 1979; Del-Rio-Navarro 2003; Fiocchi 1986; See: Summary of findings for the main comparison Summary
Garabedian 1990; Gutiérrez-Tarango 2001; Gómez-Barreto 1998; of findings table
Hauguenauer 1987; Hüls 1995; Jara-Pérez 2000; Karam-Bechara
1995; Lacomme 1985; Litzman 1999; Maestroni 1984; Motta Effect of immunostimulants (IS) on acute respiratory
1994; Paupe 1986; Pech 1987; Piquett 1986; RB10 1994; tract infections (ARTIs)
RB17 1988; RB21 1988; RB22 1990; RB24 1990; RB25 1990;
Of the 61 included studies only 35 provided data in a form suitable
Saracho-Weber 2001; Schaad 1986; Schaad 2002; Van Eygen
for inclusion in the meta-analysis for this outcome. All 35 trials
1976; Van Eygen 1979; Vautel 1993; Zagar 1988). Seven trials
compared IS with a placebo.
reported the frequency of ARTIs (Burgio 1994; Careddu 1994b;
Twenty-four out of 35 studies showed a reduction of ARTIs, both
Collet 1993; Fukuda 1999; Paupe 1991; Rutishauser 1998; Taylor
as total numbers and as a percentage reduction of ARTIs (consid-
2003). The rest of the studies did not use outcome measures rele-
ering the mean number of ARTIs in the placebo group as 100%).
vant to the prevention of respiratory infections.
In the meta-analysis the use of IS was shown to reduce the total
number of ARTIs (mean difference (MD) -1.24 95% CI -1.54
to -0.94) as well as producing a percentage change in the rate of
Other potential sources of bias ARTIs (MD -38.84%; 95% CI -46.37% to -31.31%). The total
In 29 out of the 35 included studies, the process of randomisation number of ARTIs outcome showed high heterogeneity (I2 statistic
and allocation was not described. Additionally, the disposition of = 94.0%, Chi2 test = 582.02, P < 0.00001); the use of percent-
participants and reasons for withdrawals were not reported. age change in the rate of ARTIs reduced the heterogeneity but
Funnel plots of IS effects have a considerable asymmetry, indicat- it remained very high (I2 statistic = 83.0%, Chi2 test = 195.07,
ing possible publication bias, i.e. publishing only positive results P < 0.00001). Therefore, we decided to use the random-effects
(see Assessment of reporting biases). Language bias is also possible model with MD to calculate the global effect of immunostimu-
(publication in languages other than English and publication in lants (Figure 5; Figure 6).



Figure 5. Forest plot of comparison: Any IS compared with placebo, outcome: 1.1 Mean number of ARTIs.



Figure 6. Forest plot of comparison: Any IS compared with placebo, outcome: 1.2 Per cent difference in
ARTIs.

We investigated heterogeneity by bivariate correlation as well as
by linear regression. The variables considered were total number Another potential source of heterogeneity was the type of IS. We
of children in the study, duration of the trial, mean number of decided to investigate this source of heterogeneity by performing
ARTIs in the control group versus the mean difference in number sub-analyses as follows.
of ARTIs and mean difference in the percentage of ARTIs. The 1. Including the bacterial IS studies data (and excluding the
main source of heterogeneity was the mean number of ARTIs in Saracho-Weber 2001 trial because it was the only trial with more
the control group, using the mean difference in number of ARTIs ARTIs in the IS group than in the placebo group, probably due
(linear regression model correlation - 0.672, P < 0.001). Using to a clerical mistake inverting ARTI incidences). The number of
the mean difference in the percentage of ARTIs, the source of trials was 24; total number of participants was 2154, the number
heterogeneity was related to the mean difference in the number of of IS participants was 1091 and the number of placebo
ARTIs and mean number of ARTIs in the control group (linear participants was 1063. The reduction in the total number of
regression model correlation 0.834, P < 0.001). The age of the ARTIs was MD -1.41 (95% CI -1.85 to -0.98); the reduction in
participants in each trial could be another important source of the number of ARTIs as a percentage was MD -41.21 (95% CI -
variation in the number of ARTIs, as younger children would 49.10 to -33.31).
suffer more ARTIs. Yet the age in each trial was diverse, including 2. Only bacterial IS studies data (excluding Saracho-Weber
enrolled preschool children, school-aged children and adolescents. 2001) with a total number of participants of at least 40. The
This made the exploration of this potential source of variation number of trials was 19; the total number of participants was
problematic. In general, the net reduction in the number of ARTIs 2009, the number of IS participants was 1019 and the number
was dependent on the background rate of ARTIs. of placebo participants was 990. The reduction in the total



number of ARTIs was MD -1.42 (95% CI -1.92 to -0.93); and and the reduction in the number of ARTIs as a percentage was
the reduction in the number of ARTIs as a percentage was MD - MD -36.16 (95% CI -44.51 to -27.80).
38.44 (95% CI -47.25 to -29.63). 4. Only OM-85 studies. The number of trials was nine; total
3. Only bacterial IS studies where the total number of number of participants was 852, the number of IS participants
participants was more than 40 (including D53 and OM-85 BV). was 437 and the number of placebo participants was 415. The
The number of trials was 16; the total number of participants reduction in the total number of ARTIs was MD -1.20 (95% CI
was 1811, the number of IS participants was 921 and the -1.75 to -0.66) and the reduction in the number of ARTIs as a
number of placebo participants was 890. The reduction in the percentage was MD -35.90 (95% CI -49.46 to -22.35) (Figure
total number of ARTIs was MD -1.17 (95% CI -1.56 to -0.78); 7).

Figure 7. Forest plot of comparison: OM-85 trials, outcome: 6.2 Per cent difference in ARTIs.

1. Only D53 studies. The number of trials was 11; total
number of participants was 852, the number of IS participants
was 437 and the number of placebo participants was 415. The
reduction in the total number of ARTIs was MD -1.32 (95% CI
-1.86 to -0.79); and the reduction in the number of ARTIs as a
percentage was MD -43.47 (95% CI -53.22 to -33.72) (Figure
8).

Figure 8. Forest plot of comparison: D53 trials, outcome: 7.2 Per cent difference in ARTIs.


We did not perform the sub-analyses for good quality trials as only
two out of five trials provided data as mean and SD (Del-Rio- nausea, vomiting, abdominal pain and diarrhea; and skin disorders
Navarro 2003; Gutiérrez-Tarango 2001). The selection of bacterial such as rash, urticaria and pruritus.
IS studies reduced the heterogeneity of the percentage difference The possible beneficial effects of immunostimulants in the pre-
outcome from very high (I2 statistic > 75) to moderate (I2 statistic < vention of ARTIs, cancers, AIDS/HIV infections, tuberculosis etc.
75) (Higgins 2003). However, the percentage differences and 95% have been awaited expectantly by many clinicians and medical
CI in the total number of ARTIs were similar to the overall analysis researchers. One of the main indications of registered IS is the
for all sub-analyses. The reduction of the I2 statistic when only prevention of ARTIs in children, as they are more susceptible to
the bacterial IS were analyzed confirmed that the different kinds ARTIs. IS use is common in some countries in Europe and in the
of IS were another major source of heterogeneity. The combined Americas as result of the demand to reduce the incidence of ARTIs
analysis of percentage difference outcome of D53 and OM-85 had in children.
an I2 statistic of 65%; D53 alone 55%; OM-85 alone 75%. While the use of IS is currently controversial and is viewed with
skepticism by many physicians, there have been several clinical tri-
als supporting their use as ancillary treatment and in the preven-
Adverse effects tion of ARTIs. Yet their mechanism of action and possible bene-
Twenty studies provided data on adverse events in a form suit- fits and risks are not well known. The aim of this review was to
able for inclusion in the meta-analysis. The most frequent events synthesise all the evidence currently available from trials on this
were skin and gastrointestinal effects (nausea, vomiting, abdomi- topic to enable a more robust, unbiased assessment of the role of
nal pain and diarrhea). No statistically significant difference was efficacy and safety of IS to prevent ARTIs in children.
shown in these adverse events when comparing IS with a placebo. After reviewing all available IS studies on the prevention of ARTIs
A summary of the reported safety data for each trial is noted in the in children, we empathise with the skepticism of many physicians.
outcome section of the Characteristics of included studies table. This review has a number of limitations due to the quality and
In 22 studies, authors did not report the presence or absence of reporting of the trials on this topic, the heterogeneity of the in-
adverse events (Andrianova 2003; Bánovein 1992; Clerici 1988; cluded trials and the possibility of publication bias. Few papers
Fiocchi 1986; Fiocchi 1988; Garabedian 1990; Giovannini 2000; complied with standards for methodological quality and reporting
Hauguenauer 1987; Iuldashev 1988; Longo 1988; Maestroni of clinical trials, with the majority having significant deviations
1984; Martin du Pan 1982; Piquett 1986; Prusek 1987; RB10 from these standards. In addition, the lack of detail in many of the
1994; RB17 1988; RB21 1988; RB22 1990; RB24 1990; RB25 trial publications limited the quality of this review.
1990; Saracho-Weber 2001; Sramek 1986). In eight studies, trial The most common problems with the included trials were that
authors claimed that no adverse event were observed (Chen 2004; they:
De Loore 1979; Dils 1979; Karam-Bechara 1995; Mora 2007 • did not report data on ARTIs sufficient to reproduce
Renzo 2004; Van Eygen 1976; Zagar 1988). In four studies no ad- parametric and non-parametric statistical tests or carry out meta-
verse events were observed in the immunostimulant group (Burgio analysis;
1994; Fiocchi 1989; Lacomme 1985; Schaad 1986). Five studies • did not establish the normal incidence of ARTIs in the local
reported a single case of adverse events in the IS group (Fukuda population and the number of infections in susceptible children,
1999; Gómez-Barreto 1998; Paupe 1986; Paupe 1991; Van Eygen therefore, endpoint changes were not properly established;
1979). One study reported only two adverse events (Wahl 2008). • did not identify the possible causes of recurrent ARTIs;
In three studies no adverse events related to administration of the • did not try to isolate or identify the causative pathogen;
trial medications were reported (Arroyave 1999; Jara-Pérez 2000; • did not calculate the sample size required;
Mora 2010a). • used small sample sizes;
• did not include a clinical definition of ARTI and other end
points or used non validated scales;
• misused statistical tests, especially the use of Student’s t test
for data under suspicion of non-normal distribution (SD >
DISCUSSION mean/2 or mean - SD < 0);
• under reported adverse events or did not report them at all;
Summary of main results • did not report the power of the statistical tests in trials
without significant difference;
This review shows that IS reduce the incidence of ARTIs by about
• did not report or explain the flow chart and attrition of the
40% on average (from 35 trials with a total of 4060 participants).
participants and the statistical methods did not consider the data
However, due to the poor quality of the included trials this may be
from missing participants (censored data);
an overestimate of the true effect of IS. Most of the trials reported
• published articles in low impact journals;
a low incidence of adverse events or no adverse events. The most
• included heterogeneous groups of children comprising
frequent adverse events were gastrointestinal complaints such as



infants, toddlers, schoolboys and girls and adolescents without ARTI while only 55.8% of children on echinacea had a second
considering the incidence of ARTIs in each group; infection (P = 0.01). However, this effect could not be explored
• did not control for or report on confounding factors (that is in this review.
to say, age groups, concomitant asthma or allergy, number of In all meta-analyses it is important to consider the presence and
siblings, smokers at home, birth weight, seasons during the study, possible effect of publication bias; that is, the selective publication
time and timing of attendance at daycare centre or school). of trials with positive results. The funnel plot demonstrated con-
• did not report the quality and standardisation of the herbal siderable asymmetry, indicating possible publication bias. Other
supplements (Wolsko 2005) and bacterial extracts. types of bias that could affect the funnel plot are the language bias
(publication in languages other than English and publication in
All the trials were conducted in populations of highly susceptible
non-indexed, small, local journals), poor methodological design,
children (secondary prevention) except Collet 1993 and Jara-Pérez
inadequate analysis and inadequate presentation of the results. In
2000, which studied children over-exposed to ARTIs due to the
only one study (Saracho-Weber 2001) the treated group had an
fact that they attended daycare centres and an orphanage, respec-
increase in ARTIs and, therefore, a positive difference.
tively, and Martin du Pan 1982 which included a subgroup from
The high heterogeneity limits the external validity of the analyses
daycare centres. In contrast, the echinacea trials were carried out in
with all the studies. However, when we selected only the studies
populations without a history of recurrent ARTIs (Cohen 2004;
on bacterial IS (particularly those of D53 and OM-85 BV), and
Taylor 2003).
used percentage of ARTIs, the heterogeneity could be regarded
The lack of significance of the findings of some trials could be
as acceptable. It is important to note that the size of the effects
ascribed to small sample size, duration of the intervention, season
is similar in the different sub-analyses, supporting the external
of intervention, broad participant selection criteria (mainly diverse
validity of the conclusions.
age groups) and low incidence of ARTIs in the studied population
On the basis of the asymmetry shown in the funnel plot, the het-
(caused by an over-reporting of previous ARTIs or reduction of
erogeneity of the trials and the low quality of many included trials,
incidence as children grew older).
the possibility of bias (resulting in an overestimation of the true
Thymic extracts have been withdrawn from sale in several coun-
effects of IS on ARTIs) should be considered as high. Therefore,
tries, due to the possible prion contamination and consequent
caution needs to be applied when interpreting the possible advan-
risk of bovine spongiform encephalopathy (WHO 2005). The use
tages of IS shown in this review. However, the results of this review
of levamisole is restricted because of the risk of agranulocytosis,
provide a reference to the probable effect of IS in the prevention of
neurologic disease (Symoens 1978) and leukoencephalopathy (Xu
ARTIs in children and point to the need for further clinical trials.
2009).
Further trials on IS must follow the established guidelines (Collet
The overall effect of IS was a reduction in the total number of
1992; Moher 2001; Valleron 1992), estimate sample size according
ARTIs (MD -1.24; 95% CI -1.54 to -0.94) but the individual
to realistic incidence of ARTI and control for confounding factors.
size of the effect in each trial depended on the number of ARTIs
Multivariable analysis should be used when confounding factors
in the control group. The size of the effect could seem small but
are identified. Reports of such trials must include enough data to
expressing the reduction of ARTIs as a percentage indicates a good
replicate non-parametric statistical tests (for instance, frequency
effect, about 40% (MD -38.84%; 95% CI -46.37 to -31.31). The
of ARTIs at the end point) and include statistical analysis dealing
results indicate that the reduction in the incidence of ARTIs is a
with censored data (Mahe 1999). Registration of all the protocols
real possibility but as the net effect depends on the background rate
and ongoing trials would be desirable to obtain all the possible
of ARTIs the effect would only be noticeable when the number of
outcomes. It would be necessary to conduct trials on otitis and
infections to be reduced is higher than the normal incidence for
lower ARTIs with large numbers of highly susceptible children.
a given age group. Therefore, the use of IS for the prevention of
Considering the present review, the prevention of ARTIs using
ARTIs must be limited to children with proven high susceptibility
IS may be possible. Larger clinical trials, adequately powered for
to ARTIs or over-exposed children who are over-exposed to ARTIs
important population groups, sponsored by health authorities,
because they are in daycare centres, orphanages, kindergarten or
would be desirable to establish the true effects of IS and the effect
elementary school.
of individual IS preparations.
Sub-analysis of studies with available databases (Del-Rio-Navarro
2003; Gutiérrez-Tarango 2001; Jara-Pérez 2000) have shown that
IS are not very effective in the prevention of one ARTI but are in
the prevention of recurrent infections (that is, two or three). This
may be the reason why echinacea trials fail to show protection to
the first infection but had some effect on the second infection. In
Overall completeness and applicability of
addition, the participants had no history of increased incidence
evidence
of ARTI. In a re-analysis of one study (Taylor 2003) it was found According to the funnel plots, it is possible that some trials with
that 69.2% of the children treated with placebo had a second negative results have not been published (Figure 3 and Figure 4).



Quality of the evidence reduction with OM-85 was -1.21 (95% CI -1.39 to -1.03), which
is similar to the findings in this review.
Although the global quality of the trials was poor, a group of
The results of this review do not agree with the review by Steurer-
studies comply with the quality standards (Cohen 2004; Collet
Stey (Steurer-Stey 2007) which pooled two OM-85 studies to
1993; Del-Rio-Navarro 2003; Gutiérrez-Tarango 2001; Jara-Perez
calculate the risk of fewer than three infections over six months of
2000; Taylor 2003).
follow up in children not in daycare (RR 0.82; 95% CI 0.65 to
Due to the above and the classification of the studies with the
1.02).
GRADEpro tool, the global quality of the evidence on the effect of
immunostimulants to reduce the incidence of ARTIs is regarded
as moderate. Additional research is likely to have an impact on our
confidence in the estimate of effect and may change the estimate.
Meanwhile, the evidence of the incidence of adverse events is con- AUTHORS’ CONCLUSIONS
sidered as low. Further research is very likely to have an important
impact on our confidence in the estimate of effect and is likely Implications for practice
to change the estimate, as the incidence of adverse events was not This review indicates that IS reduce the incidence of ARTIs by 40%
properly reported in most of the studies. See Summary of findings on average in susceptible children. The trials have shown benefits
for the main comparison. of IS in toddlers (two to five years), school boys (six to 12 years)
and children with a high incidence of ARTIs, for example children
attending daycare and children living in orphanages. Studies in
Potential biases in the review process healthy children are not available. Although the safety profile in the
studies was good, some IS may be unsafe. For instance, levamisole
We consider the risk of biases in the review process minimal, as
has been related to agranulocytosis and neurologic disease, and
the plausible sources of information have been consulted and the
thymic extracts introduce the risk of prion contamination and
authors and manufactures were contacted. Additionally, no exter-
therefore the risk of bovine spongiform encephalopathy.
nal funding was provided for this review.
Implications for research

Agreements and disagreements with other Further high-quality trials are required to confirm the true effect
studies or reviews of IS and individual IS preparations in the prevention of ARTIs.
We encourage national health authorities to conduct large, mul-
The present review is in agreement with a previous meta-analysis ticentre, double-blind, placebo-controlled studies to establish the
on the effect of IS showing a percent decrease in ARTIs of -42.64% precise benefits and risks for using IS to prevent ARTIs. It is nec-
(95% CI -45.19% to -40.08%) (Berber 2001). essary to conduct more studies on the number and frequency of
Another review on the effect of D53 in the incidence of ARTIs ARTIs and the physiological and immunological basis of recurrent
showed a reduction of child ear, nose and throat (ENT) infec- ARTIs.
tions of 27% to 68%, and a decrease in child ENT and bron-
chopulmonary infections of 32% to 61% compared with placebo
(Bellanti 2003). This concurs with the effect of D53 shown in this
review.
ACKNOWLEDGEMENTS
Other meta-analyses on the effect of individual IS report an effect
as a percent reduction of -31.86% (95% CI -34.32 to -29.40) for We especially acknowledged the great editorial work of Liz Doo-
D53, and a corresponding reduction of -39.28% (95% CI -52.58 ley. We thank Arturo Berber for his contribution to the protocol.
to -25.98) for OM-85 (De-La-Torre-Gonzalez 2005). Both CIs The authors also wish to thank the following people for comment-
are in agreement with those in this review. ing on the draft review of the first version: Chanpen Choprapa-
Schaad (Schaad 2010) in a meta-analysis reported that in an OM- won, Ville Peltola, Richard Shoemaker and Ludovic Reveiz; and
85 BV-treated population, 32% had three or more ARTIs in six to Anne Lyddiatt, Ville Peltola, Sree Nair and Ludovic Reveiz for
months, against 58.2% in the placebo-treated population. The the comments on the updated draft review.


Immunostimulants for preventing respiratory infections in children

Immunostimulants for preventing respiratory infections in children

Recommended

Recommended

More Related Content

Viewers also liked

Viewers also liked (10)

Similar to Immunostimulants for preventing respiratory infections in children

Similar to Immunostimulants for preventing respiratory infections in children (20)

Immunostimulants for preventing respiratory infections in children