Grading of Recommendations Assessment, Development and Evaluation (GRADE) findings

S Thangaratinam; E Rogozińska; K Jolly; S Glinkowski; W Duda; E Borowiack; T Roseboom; J Tomlinson; J Walczak; R Kunz; BW Mol; A Coomarasamy; KS Khan

5Grading of Recommendations Assessment, Development and Evaluation (GRADE) findings

Prioritisation of outcomes

The primary outcomes were weight-related outcomes. There were numerous secondary outcomes. These were ranked through a two-iteration Delphi survey.

First iteration

A total of 19 clinicians (19/20, 95%) completed the questionnaire. Five maternal outcomes – GDM, pre-eclampsia and pregnancy-induced hypertension, caesarean section, thromboembolism and admission to the HDU/intensive therapy unit (ITU) – had a median score of ≥ 8 with an IQR of ≤ 2. The six fetal outcomes that were scored in a similar fashion were SGA infants, intrauterine death, admission to NICU, shoulder dystocia, birth trauma and long-term neurological sequelae. In addition to the outcomes provided, the panel considered breastfeeding, back pain, threatened miscarriage, failed instrumental delivery, maternal coronary artery disease, maternal non-infective respiratory distress, cord abnormalities and long-term metabolic sequelae in the infant to be relevant to the question posed. These outcomes were added to the initial outcomes and sent for scoring for importance in the second round.

Second iteration

A total of 16 panellists (16/19, 84%) participated in the second round of the survey. For maternal outcomes there was evidence of consensus for GDM, thromboembolism and admission to HDU/ITU, as reflected in the median scores of 8 and a fall in IQR from the first round score. Preeclampsia continued to be considered as a critically important outcome, with a median score of > 8, although there was an increase in the IQR from 1.5 to 2. Induction of labour scored a median of 8 and was included in the final list of outcomes. Caesarean section as an outcome scored lower (median 7) than in the first round.

For fetal outcomes there was consistency in the ranking, with median scores of > 8 and IQRs of ≤ 1.25 for birth trauma, intrauterine death, admission to NICU and shoulder dystocia. All of the selected fetal outcomes consistently demonstrated a narrowing of the IQR scores in the second round, demonstrating consensus between the participants. The ten outcomes considered to be critical to patient care are provided in Box 2. The scores for the outcomes in the two rounds of the Delphi survey are provided in Appendix 11.

BOX 2

Delphi panel list of outcomes of critical importance in the management of maternal weight in pregnancy. GDM Pre-eclampsia/gestational hypertension

Grading of evidence for the effectiveness and adverse effects of interventions

The grading of the evidence for the primary outcomes related to maternal and fetal weight commissioned by the HTA programme and the outcomes considered to be critically important for patient management are summarised graphically in Figure 37. This two-dimensional chart plots five variables represented by equiangular spokes, which represent the quality domains used in evidence grading for each comparison–outcome pair. For each of the spokes, the length represents the magnitude of the quality, ranging from very low at the centre of the plot to high at its maximum length.

FIGURE 37

Graphic display of the evidence quality for the effect of various interventions on weight-related and clinically important outcomes.

Details of the quality assessment are provided in Appendix 12. The overall strength of evidence for weight gain in pregnancy and birthweight was moderate for all interventions considered together. The strength of evidence for all interventions together was moderate for shoulder dystocia and high for SGA infants. The quality of the pooled evidence for all interventions was moderate for gestational hypertension in obese and overweight women and intrauterine death, and low for reduction in pre-eclampsia and birth trauma. The trend in reduction of GDM was graded low (Table 8). Although thromboembolism, maternal admission to HDU/ITU and long-term neurological sequelae to the fetus were considered to be critically important to the clinicians, we did not identify relevant evidence for these outcomes. Dietary interventions in pregnancy were graded moderate to high for the important outcomes more often than the other interventions (see Appendix 13).

TABLE 8

The GRADE profile of the RCTs on the effects of weight management interventions in pregnancy on the primary and clinically important outcomes.

The quality of the evidence for adverse outcomes for studies reporting diet and physical activity in pregnancy is provided in Table 9. The strength of evidence was very low for all of the outcomes evaluated for dietary intervention. Poor quality of evidence was also observed for physical activity interventions in pregnancy.

TABLE 9

GRADE profile for adverse effects due to diet and physical activity in pregnancy.

Summary

The Delphi survey prioritised outcomes that were considered to be critical in the management of women in pregnancy. The evidence quality on the primary outcomes related to weight, maternal weight gain in pregnancy and birthweight was graded as moderate. The strength of evidence was low for secondary outcomes such as pre-eclampsia, GDM, gestational hypertension and caesarean section and low to high for preterm birth, induction of labour, shoulder dystocia, birth trauma, incidence of SGA and LGA infants and intrauterine death for all interventions. The strength of evidence for adverse outcomes due to diet and physical activity was mostly very low reflecting the paucity of evidence in this area.

The weight-related outcomes were regarded as critical in the HTA commissioning brief (HTA No. 09/27/06) for an evaluation of the reduction or prevention of obesity in pregnancy. In addition to the large benefits observed with dietary intervention, the strength of evidence for this intervention was also rated better than that for the other interventions. The evidence for gestational weight gain was of moderate quality for dietary interventions and low for the physical activity and mixed approach interventions. For subgroups of overweight women and obese women the strength of evidence was low to very low for all three interventions. This was a result of the imprecision in the estimates and incomplete reporting of the outcome data. The quality of evidence for the incidence of SGA infants, which showed no significant differences between the intervention and control groups, was moderate to high for all of the interventions. This finding is reassuring to an extent as it negates the perceived risks of interventions for the growth of the fetus.

The evidence quality for reduction in the rate of pre-eclampsia was moderate for dietary intervention, which showed the largest reduction in risk. In the subgroups of obese and overweight women the beneficial effect of dietary intervention in reducing pre-eclampsia scored a moderate-to-high grade for the quality of evidence. Overall, there was moderate-quality evidence that weight management interventions reduce the risks of shoulder dystocia, with the potential to reduce associated morbidity and mortality. The strength of evidence was low for the trend towards a reduction in the incidence of GDM. It is possible that a different panel may have identified a different group of clinically important outcomes.

The graphic display has captured the quality of the evidence for many comparisons and outcomes simultaneously in one diagram making it possible to comprehend large numbers of data in one glance. The diagram, once understood, allows for appraisal of key issues concerning risk of bias, heterogeneity, directness of evidence in relation to the question, and precision of results. This critical appraisal alters the trust that we can place in the evidence collated for decision-making.

The GRADE profile findings are limited because of the paucity of evidence for some important outcomes such as thromboembolism, maternal admission to HDU/ITU, long-term neurological sequelae and more than one perinatal complication. Further research is likely to have an important impact on the confidence of our estimate and is likely to change the estimate. We have refrained from assessing the quality of evidence across outcomes as it is in the domain of the guideline developers. As systematic reviewers we have limited ourselves to the GRADE profiling of the important outcomes.

Publication Details

Copyright

Included under terms of UK Non-commercial Government License.

Publisher

NIHR Journals Library, Southampton (UK)

NLM Citation

Thangaratinam S, Rogozińska E, Jolly K, et al. Interventions to Reduce or Prevent Obesity in Pregnant Women: A Systematic Review. Southampton (UK): NIHR Journals Library; 2012 Jul. (Health Technology Assessment, No. 16.31.) 5, Grading of Recommendations Assessment, Development and Evaluation (GRADE) findings.

BOX 2Delphi panel list of outcomes of critical importance in the management of maternal weight in pregnancy

GDM

Pre-eclampsia/gestational hypertension

Admission to HDU/ITU

Thromboembolism

Induction of labour

SGA infants

Shoulder dystocia

Birth trauma

Admission to NICU

Long-term neurological sequelae

FIGURE 37Graphic display of the evidence quality for the effect of various interventions on weight-related and clinically important outcomes

TABLE 8The GRADE profile of the RCTs on the effects of weight management interventions in pregnancy on the primary and clinically important outcomes

Outcomes	Illustrative comparative risks^a (95% CI)		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)^b

	Assumed risk, control	Corresponding risk, all weight management interventions
Gestational weight gain (kg)		The mean gestational weight gain (kg) in the intervention groups was 0.94 kg lower (1.57 kg to 0.3 kg lower)		4595 (30)	⊕⊕⊕⊖ Moderate^c^,^d^,^e
Birthweight (kg)		The mean birthweight (kg) in the intervention groups was 0.07 kg lower (0.14 kg to 0.01 kg lower)		4573 (28)	⊕⊕⊕⊖ Moderate^c^,^d^,^f
LGA	157 per 1000	115 per 1000 (85 to 155)	RR 0.73 (0.54 to 0.99)	3021 (12)	⊕⊕⊕⊖ Moderate^c^,^d^,^g
SGA	71 per 1000	70 per 1000 (54 to 92)	RR 0.99 (0.76 to 1.29)	2901 (8)	⊕⊕⊕⊕ High^h
Pre-eclampsia	103 per 1000	76 per 1000 (61 to 95)	RR 0.74 (0.59 to 0.92)	3072 (10)	⊕⊕⊖⊖ Low^h^,ⁱ^,^j
Gestational hypertension	134 per 1000	103 per 1000 (72 to 147)	RR 0.77 (0.54 to 1.1)	791 (6)	⊕⊕⊖⊖ Low^k
GDM	112 per 1000	80 per 1000 (49 to 127)	RR 0.71 (0.44 to 1.13)	675 (5)	⊕⊕⊖⊖ Low^h^,^l^,^m
Preterm birth	82 per 1000	62 per 1000 (46 to 84)	RR 0.76 (0.56 to 1.02)	2198 (11)	⊕⊕⊕⊖ Moderate^c^,^d^,^g
Caesarean section	327 per 1000	304 per 1000 (278 to 337)	RR 0.93 (0.85 to 1.03)	3312 (14)	⊕⊕⊖⊖ Low^c^,^d^,^g^,^m
Induction of labour	295 per 1000	330 per 1000 (295 to 372)	RR 1.12 (1.0 to 1.26)	2362 (5)	⊕⊕⊕⊖ Moderate^c^,^d^,^g
Post-partum haemorrhage	59 per 1000	53 per 1000 (24 to 84)	RR 0.90 (0.57 to 1.43)	1232 (2)	⊕⊕⊖⊖ Low^g,ⁿ
Intrauterine death	9 per 1000	1 per 1000 (0 to 11)	RR 0.15 (0.02 to 1.2)	1320 (2)	⊕⊕⊕⊖ Moderate^h
Admission to NICU	382 per 1000	374 per 1000 (252 to 562)	RR 0.98 (0.66 to 1.47)	1962 (2)	⊕⊖⊖⊖ Very low^c^,^d^,^g^,ⁱ^,^l
Shoulder dystocia	33 per 1000	13 per 1000 (7 to 23)	RR 0.39 (0.22 to 0.7)	2317 (4)	⊕⊕⊕⊖ Moderate^g
Birth trauma	9 per 1000	3 per 1000 (1 to 11)	RR 0.36 (0.11 to 1.23)	1961 (2)	⊕⊕⊖⊖ Low^g^,^h
Neonatal hypoglycaemia	103 per 1000	110 per 1000 (88 to 139)	RR 1.07 (0.85 to 1.35)	2146 (5)	⊕⊖⊖⊖ Very low^g^,^l^,^m

a: Poor information about allocation concealment which was assess as not strongly significant.

b: Poor information about blinding of subjective outcomes which was assessed as not strongly significant.

c: High risk of bias regarding incompleteness of outcome data addressed and selective reporting.

d: High risk of bias regarding incompleteness of outcome data addressed.

e: Women with gestational diabetes.

f: Allocation concealment not clear but not considered to be necessary for downgrading.

g: Qualitative difference in the summary estimate.

h: Significant dubgroup effect observed for women with gestational diabetes.

i: Heterogeneity I2 = 48%.

j: Wide confidence interval crossing line of no effect.

k: Slight skew in funnel plot for given outcome.

l: Difficult to interpret as only two studies.

m: Evidence only for one group of interventions.

TABLE 9GRADE profile for adverse effects due to diet and physical activity in pregnancy

TABLE 9aGRADE profile for adverse effects due to diet and physical activity in pregnancy: diet in pregnancy

Outcomes	Illustrative comparative risks^a (95% CI)		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)^b

	Assumed risk, control	Corresponding risk, diet
NTD	See comment ^o	See comment ^o	Not estimable	0 (2)	⊕⊖⊖⊖ Very low^c^,^d^,^e
Coronary heart disease: long-term outcome in children	32 per 1000	90 per 1000 (35 to 209)^f	OR 3 (1.1 to 8.0)	508 (1)	⊕⊖⊖⊖ Very low^e^,^g^,^h
Metabolic syndrome: long-term outcome in children	1 per 1000	1 per 1000 (1 to 2)ⁱ	OR 1.2 (0.9 to 1.7)	59,317 (1)	⊕⊖⊖⊖ Very low^g^,^h
Hypertension: long-term outcome in children	571 per 1000	651 per 1000 (576 to 720)	OR 1.4 (1.02 to 1.93)	638 (1)	⊕⊖⊖⊖ Very low^h^,^j
Antisocial personality disorder: long-term outcome in children	1 per 1000	2 per 1000 (1 to 3)ⁱ	OR 2.0 (1.2 to 3.3)	59,317 (1)	⊕⊖⊖⊖ Very low^e^,^g^,^h
Dyslipidaemia	289 per 1000	279 per 1000 (199 to 353)	OR 0.95 (0.61 to 1.34)	638 (1)	⊕⊖⊖⊖ Very low^h^,^j^,^k
Obesity: in adulthood: long-term outcome in children	13 per 1000^l	8 per 1000 (6 to 11)^l	OR 0.62 (0.45 to 0.85)	17,400 (1)	⊕⊖⊖⊖ Very low^h^,^m
Obesity: in adulthood: long-term outcome in children	14 per 1000ⁿ	27 per 1000 (22 to 33)ⁿ	OR 1.94 (1.55 to 2.43)	20,200 (1)	⊕⊖⊖⊖ Very low^h^,^m
IGT: long-term outcome in children	85 per 1000	94 per 1000	OR 1.12 (0.54 to 2.32)	357 (1)	⊕⊖⊖⊖ Very low^g^,^h

: IGT, impaired glucose tolerance; max., maximum.

a: 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% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

b: 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.

c: Poor information about definition of controls, ascertainment of exposure and non-response rate in Yazdy 2010¹⁴³ study.

d: No explanation was provided.

e: OR > 2.

f: Exposed early.

g: Observational study, 5 points (9 max.) in NOS questionnaire.

h: Singleton men and women born between January 1945 and March 1946 whose mothers were exposed or not to the Dutch famine during pregnancy.

i: During first, second or third trimester.

j: Observational study, 6 points (9 max.) in NOS questionnaire.

k: Wide CI.

l: During third trimester of pregnancy.

m: Observational study, 4 points (9 max.) in NOS questionnaire.

n: During first and second trimester of pregnancy.

o: Data heterogeneous and not suitable for pooling of estimates.

TABLE 9bGRADE profile for adverse effects due to diet and physical activity in pregnancy: physical activity in pregnancy

Outcomes	Illustrative comparative risks^a (95% CI)		Relative effect (95% CI)	No. of participants (studies)	Quality of the evidence (GRADE)^b

	Assumed risk, control	Corresponding risk, physical activity
Cord abnormalities	83 per 1000	37 per 1000 (17 to 82)^c	OR 0.43 (0.19 to 0.99)	455 (3)	⊕⊖⊖⊖ Very low^d
Stimulation for abnormal labour pattern	205 per 1000	115 per 1000 (43 to 303)	RR 0.56 (0.21 to 1.48)	131 (1)	⊕⊖⊖⊖ Very low^d^,^e
Meconium in amniotic fluid	170 per 1000	105 per 1000 (34 to 323)	RR 0.62 (0.2 to 1.9)	85 (1)	⊕⊕⊖⊖ Low^e^,^f^,^g^,^h
Abnormal fetal heart rate	250 per 1000	138 per 1000 (60 to 286)	OR 0.48 (0.19 to 1.2)	131 (1)	⊕⊖⊖⊖ Very low^d^,^e
Nuchal cord	545 per 1000	264 per 1000 (144 to 434)	OR 0.3 (0.14 to 0.64)	131 (1)	⊕⊖⊖⊖ Very low^d
Threatened abortion	91 per 1000	48 per 1000 (3 to 470)	OR 0.5 (0.03 to 8.85)	32 (1)	⊕⊖⊖⊖ Very low^e^,ⁱ
Failure to progress with oxytocin augmentation	273 per 1000	142 per 1000 (26 to 503)	OR 0.44 (0.07 to 2.7)	32 (1)	⊕⊖⊖⊖ Very low^e^,ⁱ
Chorioamnionitis	26 per 1000	0 per 1000	OR 3.69 (0.15 to 88.13)	85 (1)	⊕⊕⊖⊖ Low^e^,^f^,^g^,^h
Maternal anaemia	182 per 1000	143 per 1000 (24 to 541)	OR 0.75 (0.11 to 5.3)	32 (1)	⊕⊖⊖⊖ Very low^e^,ⁱ
Maternal sepsis	91 per 1000	16 per 1000 (1 to 303)	OR 0.16 (0.01 to 4.35)	32 (1)	⊕⊖⊖⊖ Very low^e^,ⁱ
Uterine atony	85 per 1000	79 per 1000 (19 to 331)	RR 0.93 (0.22 to 3.89)	85 (1)	⊕⊕⊖⊖ Low^e^,^f^,^g^,^h

: max., maximum.

a: 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% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

b: 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.

c: Exercise (heavy).

d: Observational study, 6 points (9 max.) in NOS questionnaire.

e: Wide CI.

f: Limited information about allocation concealment, assessed as not strongly significant.

g: Limited information about blinding of subjective outcomes, assessed as not strongly significant.

h: Limited information about adequate sequence generation, assessed as not strongly significant.

i: Observational study, 4 points (9 max.) in NOS questionnaire.