Appendix 1: GRADE tables on efficacy evidence for tocolytic therapy (excluding information on unimportant outcomes*).
Evidence quality on tocolytic efficacy compared to placebo or no treatment
(corresponding to Figure 1)
Tables A-D
Evidence quality on tocolytic efficacy compared to Beta agonists
(corresponding to Figure 2)
Tables E-I
* Tables for all outcomes and comparisons available from authors on request. References to all individual studies included in all tables also available from authors on request. References to individual studies included in this appendix are lisetd at the end of the GRADE tables.
Table A. Atosiban vs placebo
No of patients / Effect / Quality /
No of studies / Design / Limitations / Inconsistency / Indirectness / Imprecision / Other considerations / Atosiban / Placebo / Relative
(95% CI) / Absolute /
Perinatal morbidity (Based on neonatal admission to intensive care unit)1
1 / randomised trials / serious1,2 / no serious inconsistency / no serious indirectness3 / no serious imprecision / none / 115/274 (42%) / 110/286 (38.5%) / RR 1.09 (0.89 to 1.34) / 35 more per 1000 (from 42 fewer to 131 more) / ÅÅÅO
MODERATE / 1
Safety to mother (Based on maternal drug reaction requiring treatment cessation) 1, 2
2 / randomised trials / serious / serious / Serious / Not serious / none / 40/306 (13.7%) / 10/307 (3.3%) / RR 4.02 (2.05 to 7.85) / 98 more per 1000 (from 34 more to 223 more) / ÅOOO
VERY LOW / 2
Perinatal mortality 1
1 / randomised trials / serious1,2 / no serious inconsistency / no serious indirectness3 / no serious imprecision / none / 11/288 (3.8%) / 5/295 (1.7%) / RR 2.25 (0.79 to 6.4) / 21 more per 1000 (from 4 fewer to 92 more) / ÅÅÅO
MODERATE / 3
Undelivered at 24 hours after initiation of treatment 1
1 / randomised trials / no serious limitations1,2 / no serious inconsistency / no serious indirectness3 / no serious imprecision / none / 179/246 (72.8%) / 148/255 (58%) / RR 1.25 (1.1 to 1.43) 4 / 145 more per 1000 (from 58 more to 250 more) / ÅÅÅÅ
HIGH / 5
Undelivered at 48 hours after initiation of treatment 1
1 / randomised trials / no serious limitations1,2 / no serious inconsistency / no serious indirectness3 / no serious imprecision / none / 165/246 (67.1%) / 142/255 (55.7%) / RR 1.20 (1.05 to 1.39) 4 / 111 more per 1000 (from 28 more to 217 more) / ÅÅÅÅ
HIGH / 6
Stop birth before 37 weeks of gestation 1
1 / randomised trials / serious1,2 / no serious inconsistency / no serious indirectness3 / no serious imprecision / none / 144/246 (58.5%) / 128/255 (50.2%) / RR 1.17 (0.99 to 1.37) / 85 more per 1000 (from 5 fewer to 186 more) / ÅÅÅO
MODERATE / 7
Footnotes:
1 Data not analyzed according to intention-to-treat principle.
2 There was an imbalance, despite randomization, at study entry of women before 26 weeks of gestation between atosiban and placebo groups (24/246 (10%) versus 13/255 (5%) respectively), with fewer women in the atosiban group above 32 weeks of gestation compared with placebo (96/246 versus 116/255). This biases in favor of null effect.
3 Eligibility criteria for participation in the study were quite different from those in Summary of Product Characteristics. This limitation could lead to overestimation or underestimation of the effect of the assessed interventions.
4 Statistically significant result in favour of atosiban.
* Ranking of importance undertaken through formal survey of practitioners asking them to provide responses on a scale anchored between critical at one extreme and not at all important at the other.
Table B. Beta agonists vs placebo
No of patients / Effect / Quality /
No of studies / Design / Limitations / Inconsistency / Indirectness / Imprecision / Other considerations / Beta agonists / Placebo / Relative
(95% CI) / Absolute /
Perinatal death3-13
11 / randomised trials / serious1,2 / serious3 / serious4,5 / no serious imprecision / None / 16/712 (2.2%) / 20/620 (3.2%) / RR 0.84 (0.46 to 1.55) / 5 fewer per 1000 (from 17 fewer to 18 more) / ÅOOO
VERY LOW / 3
Undelivered at 48 hours after initiation of treatment 3,4,6-13
10 / randomised trials / serious1,6 / serious3 / no serious indirectness5 / no serious imprecision / None / 151/652 (23.2%) / 218/557 (39.1%) / RR 0.63 (0.53 to 0.75)7 / 145 fewer per 1000 (from 98 fewer to 184 fewer) / ÅÅOO
LOW / 6
Stop birth before 37 weeks of gestation3,4,6-13
10 / randomised trials / serious1,8 / serious3 / serious4,5 / no serious imprecision / None / 404/654 (61.8%) / 383/558 (68.6%) / RR 0.95 (0.88 to 1.03) / 34 fewer per 1000 (from 82 fewer to 21 more) / ÅOOO
VERY LOW / 7
Footnotes:
1 Information about power calculation was not presented, so it was impossible to assess whether sample size was adequate.
2 From 11 publications 8 are stated to be double-blind and 1 to be single-blind. Only in 3 publications information about proper method of randomisation and allocation concealment is placed. In 7 publications number of patients was small (less than 20 in each treatment group). From 11 studies 4 were not estimable.
3 It should be noted that differences in the criteria for inclusion and exclusion from the studies (eg. membranes status, multiple pregnancy, gestational age) and the different treatment regimens or the possibility of use different alternative tocolytic agents may be the main causes of inconsistency.
4 It should be noticed that metaanalysis was based on clinical trials where maintenance therapy was employed. We decided to decrease the quality rating, because the results of such a metaanalysis are less useful in informing about the effectiveness of initial (loading) regimens.
5 In one study patients at the beginning received 10 mg of diazepam intramuscular, which can be harmful for the infant. Bed rest and sedation were also part of treatment.
6 From 10 publications 7 are stated to be double-blind and 1 to be single-blind. Only in 3 publications information about proper method of randomisation and allocation concealment are placed. In 7 publications number of patients was small (less than 20 in each treatment group).
7 Statistically significant result in favour of Beta agonists.
8 From 10 publications 7 are stated to be double-blind and 1 to be single-blind. Only in 3 publications information about proper method of randomisation and allocation concealment was placed. In 7 publications number of patients was small (less or equal 20).
*Ranking of importance undertaken through formal survey of practitioners asking them to provide responses on a scale anchored between critical at one extreme and not at all important at the other.
Table C. Indomethacin vs placebo
No of patients / Effect / Quality /
No of studies / Design / Limitations / Inconsistency / Indirectness / Imprecision / Other considerations / Indomethacin / Placebo / Relative
(95% CI) / Absolute /
Perinatal morbidity (Based on neonatal admission to intensive care unit) 14
1 / randomised trials / no serious limitations1,2 / no serious inconsistency / no serious indirectness3 / serious4 / None / 13/19 (68.4%) / 17/20 (85%) / RR 0.80 (0.56 to 1.15) / 170 fewer per 1000 (from 374 fewer to 127 more) / ÅÅÅO
MODERATE / 1
Perinatal mortality 14,15,16
3 / randomised trials / no serious limitations1,2,5 / serious6 / serious3,7 / no serious imprecision / None / 4/53 (7.5%) / 5/53 (9.4%) / RR 0.80 (0.25 to 2.58) / 19 fewer per 1000 (from 71 fewer to 149 more) / ÅÅOO
LOW / 3
Undelivered at 48 hours after initiation of treatment 14, 16
2 / randomised trials / no serious limitations1,2,5 / no serious inconsistency / no serious indirectness3 / no serious imprecision / None / 4/34 (11.8%) / 22/36 (61.1%) / RR 0.19 (0.07 to 0.51)8 / 495 fewer per 1000 (from 299 fewer to 568 fewer) / ÅÅÅÅ
HIGH / 6
Stop birth before 37 weeks of gestation 16
1 / randomised trials / no serious limitations1,9 / no serious inconsistency / serious7 / no serious imprecision / None / 3/18 (16.7%) / 14/18 (77.8%) / RR 0.21 (0.07 to 0.62)8 / 614 fewer per 1000 (from 296 fewer to 723 fewer) / ÅÅÅO
MODERATE / 7
Footnotes:
1 Number of patients in each treatment group was small (less than 20).
2 Information about sample size calculation was presented only in one study, but the study sample was obviously small for key outcomes.
3 Some patients use tocolytic agents like MgS04 and/or ritodrine in 48h before study entry.
4 Downgraded because of non significant RR result; wide confidence interval; obvious differences in percentage of cases in each treatment groups; small sample of patients.
5 One publication included incomplete information about method of randomisation and allocation concealment.
6 It should be noted that differences in the criteria for inclusion and exclusion from the studies (eg. multiple pregnancy, gestational age) and the different treatment regimens or the possibility of use different alternative tocolytic agents may be the main causes of inconsistency.
7 In study Zuckerman 1984 dosage of indomethacin was higher (100mg rectally) than recommended 50mg. Additionally 100mg suppository if required was possible.
8 Statistically significant result in favor of indomethacin.
9 Publication included incomplete information about method of randomisation and allocation concealment. Information about power calculations was not presented, so it was impossible to assess whether sample size was adequate.
*Ranking of importance undertaken through formal survey of practitioners asking them to provide responses on a scale anchored between critical at one extreme and not at all important at the other.
Table D. Nitric oxide donors vs placebo/no treatment
No of patients / Effect / Quality /
No of studies / Design / Limitations / Inconsistency / Indirectness / Imprecision / Other considerations / Nitric oxide donors / Placebo/no treatment / Relative
(95% CI) / Absolute /
Perinatal mortality17
1 / randomised trials / no serious limitations / no serious inconsistency / no serious indirectness / No serious imprecison / None / 0/74 (0%) / 3/76 (3.9%) / RR 0.15 (0.01 to 2.79) / - / ÅÅÅÅ
HIGH / 1
Undelivered at 48 hours after initiation of treatment 18
1 / randomised trials / no serious limitations / no serious inconsistency / no serious indirectness / Very serious2 / None / 6/17 (35.3%) / 10/16 (62.5%) / RR 0.56 (0.27 to 1.19) / 275 fewer per 1000 (from 456 fewer to 119 more) / ÅÅOO
LOW / 6
Stop birth before 37 weeks of gestation17
1 / randomised trials / no serious limitations / no serious inconsistency / no serious indirectness / No serious imprecison / None / 38/74 (51.4%) / 42/79 (53.2%) / RR 0.97 (0.71 to 1.31) / - / ÅÅÅÅ
HIGH / 7
Footnotes:
1 Statistically significant result in favour of placebo.
2 This was a pilot study with small number of patients. The confidence interval for this endpoint was not particularly wide, but we decided to downgrade the quality of evidence because the differences in percentages of events between groups was without statistical significance.
* Ranking of importance undertaken through formal survey of practitioners asking them to provide responses on a scale anchored between critical at one extreme and not at all important at the other.
Table E. Atosiban vs Beta agonists
No of patients / Effect / Quality
No of studies / Design / Limitations / Inconsistency / Indirectness / Imprecision / Other considerations / atosiban / Beta agonists / Relative
(95% CI) / Absolute
Perinatal morbidity (Based on neonatal admission to intensive care unit19-22
4 / randomised trials / serious1,2,3 / serious4 / no serious indirectness5 / no serious imprecision / None / 142/486 (29.2%) / 142/494 (28.7%) / RR 1.02 (0.84 to 1.24) / 6 more per 1000 (from 46 fewer to 69 more) / ÅÅOO
LOW / 1
Safety to mother (Based on maternal drug reaction requiring treatment cessation) 19-23
5 / randomised trials / very serious1,2,3,6 / serious4 / serious5,7,8 / no serious imprecision / None / 4/667 (0.6%) / )88/495 (17.8%) / RR 0.03 (0.01 to 0.09)9 / 167 fewer per 1000 (from 155 fewer to 171 fewer) / ÅOOO
VERY LOW / 2
perinatal mortality 19-22
4 / randomised trials / serious1,2,3 / serious4 / no serious indirectness5 / no serious imprecision / None / 7/486 (1.4%) / 12/494 (2.4%) / RR 0.59 (0.24 to 1.49) / 10 fewer per 1000 (from 18 fewer to 12 more) / ÅÅOO
LOW / 3
Undelivered at 24 hours of initiation of treatment22
1 / randomised trials / serious2 / no serious inconsistency / no serious indirectness / no serious imprecision / None / 59/63 (93.7%) / 60/63 (95.2%) / RR 0.98 (0.90 to 1.07) / 16 more per 1000 (from 33 fewer to 225 more) / ÅÅÅO
MODERATE / 5
Undelivered at 48 hours after initiation of treatments19-21
3 / randomised trials / serious1,3 / serious4 / no serious indirectness5 / no serious imprecision / None / 187/360 (51.9%) / 260/372 (69.9%) / RR 1.07 (0 to 1.17) / 332 more per 1000 (from 236 fewer to 384 more) / ÅÅÅO
MODERATE / 6
Stop birth before 37 weeks of gestation20
1 / randomised trials / no serious limitations1 / no serious inconsistency / no serious indirectness / no serious imprecision / None / 60/115 (52.2%) / 75/129 (58.1%) / RR 0.90 (0.71 to 1.13) / 58 fewer per 1000 (from 169 fewer to 76 more) / ÅÅÅÅ
HIGH / 7
Footnotes: