2016 Section 5 Green Book

dence.

The following system was used

to assign gradewhen looking at the over-

all evidence for each outcome: high-

grade evidence required a minimum of

2 or more trials with low risk of bias, and

at least 1 strong magnitude of effect in

the context of largely consistent overall

evidence. Moderate-grade evidence re-

quired 1 or more trials with low risk of

bias or strong magnitude of effect, or 1

trial with low risk of bias or moderate

magnitude plus 1 trial withmedium risk

of bias or strong magnitude. Evidence

was low grade if it did not meet any of

these categories. Insufficient evidence

was assigned if there were no relevant

trials. The team reviewed and came to

consensus on the grades. The evidence

regarding indirect outcome measures

(pulmonary function test results and

provocation tests) was not graded.

A meta-analysis was not performed

due to the extreme heterogeneity of the

included studies. Therewas awide range

of clinical diversity in the types of par-

ticipants, their allergies, allergens treated,

and geographic treatment locations. The

extreme variability of the publishedRCTs

in dosing and treatment schedules con-

foundsmeta-analysis.Methodological di-

versity was evident in the study de-

signs, quality, and systems for measured

outcomes. The statistical reporting was

incomplete in most studies and infor-

mation such as confidence intervals was

rarely reported.

To reduce the risk of publication bias

in our results, we searched public reg-

istries of clinical trials and requested sci-

entific information packets from rel-

evant pharmaceutical companies to

search for unpublished trials. Because

measures of variance were not reported

inmost of the included studies, we could

not produce a meaningful funnel plot to

look for evidence of publication bias.

RESULTS

We identified 8156 potentially relevant

citations. After applying exclusions and

triage toother topics, 63RCTs remained

for this systematic review (

IGURE

). We

identified63RCTs testingsublingual im-

munotherapywith5131participants in-

cluded. Participants’ ages ranged from

4 to 74 years. Twenty-six studies (41%)

enrolled adults only, 17 (27%) included

both adults and children,

8-24

and 20

(32%) exclusivelystudiedchildren( 18

years of age; n=1814 patients).

25-44

Thestudycomparatorgroupswerepla-

cebo (46 studies; 73%), another sublin-

gualinterventionwithoutaplacebogroup

(9 studies; 14%), andconventional treat-

ment(pharmacotherapy)withoutplacebo

Figure.

Flow Diagram of Sublingual Immunotherapy Studies

63 Randomized controlled trails included in

systematic review (N=5131 participants)

184 Potentially relevant articles identified

for data abstraction

1827 Potentially relevant articles identified

for article review

8156 Potentially relevant articles identified

for title and abstract review

17222 Citations identified from electronic

database searches

9650 EMBASE

6633 MEDLINE

840 Cochrane Central Register

of Controlled Trials

99 LILACS

121 Excluded

85 No sublingual immunotherapy

9 No diagnosis

7 No dose specified

7 Data not abstractable

6 Not a randomized controlled trial

5 No comparison group

5 No outcomes

1 Therapy or related therapy not approved in

the United States

77 Subcutaneous immunotherapy

8 Sublingual immunotherapy vs subcutaneous

immunotherapy

1643 Excluded

595 Did not meet all of the inclusion criteria

370 Did not apply to any of the key questions

260 Wrong study design

228 No original data

185 Other reasons

177 Therapy or related therapy not approved in

the United States

66 Published in foreign language (non-English)

51 Abstract only

46 No specific immunotherapy

43 Library unable to locate article

21 Part of another study

6329 Excluded

3677 Did not apply to any of the key questions

1871 No specific immunotherapy

1337 No original data

589 Therapy or related therapy not approved in

the United States

141 No patients with allergic rhinoconjunctivitis

81 Study evaluated outcomes in animals only

42 No comparison group and no report of harms

23 Number of patients in the study receiving

treatment was 6 or fewer

9066 Excluded (duplicates)

Articles were excluded by 2 reviewers at this level so this number is exceeded by the total for all of the ex-

clusions listed.

The possible other reasons included control group is a healthy population, routes of administration not in-

cluded (eg, oral, nasal, lymph node), abandoned interventions, outcomes not reported, no comparator group,

continued medical education reports, editorials or reviews, studies about mechanism of action, other allergies

(eg, food, aspirin), study in animals or in vitro, or fewer than 6 patients per treatment group.

SUBLINGUAL IMMUNOTHERAPY FOR RHINOCONJUNCTIVITIS AND ASTHMA

JAMA,

March

27,

2013—Vol

309, No.

Corrected on July 29, 2013

185