GRADE evaluation

Question: Should ablation vs hysterectomy be used for HMB? Bibliography: Lethaby 2010 (Cochrane review)Endometrial resection and ablation versus hysterectomy for heavy menstrual bleeding
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
							With Hysterectomy	With Ablation		Risk with Hysterectomy	Risk difference with Ablation (95% CI)
post operative recovery (measured with: Time to return to normal activities; range of scores: 0-40; Better indicated by lower values)
632 (4 studies) 2 years	no serious risk of bias	serious¹	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE¹ due to inconsistency	287	345	MD -11.3		The mean post operative recovery in the intervention groups was -11.33 lower (-19.79 to -2.87 lower)

¹ I2 is 97%

Question: Should first generation ablation vs second generation ablation be used for HMB? Bibliography: Batthacharya 2012Hysterectomy, endometrial ablation and Mirena for heavy menstrual bleeding: a systematic review of clinical effectiveness and costeffectiveness analysis Or Cochrane reviewLethaby 2009: endometrial resection/ablation techniques for heavy menstrual bleeding
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
							With Second generation ablation	With First generation ablation		Risk with Second generation ablation	Risk difference with First generation ablation (95% CI)
amenorrhoea after 1 year (assessed with: PBAC chart/Amenorrhoea rate) Batthacharya 2012
2180 (13 studies) 1 years	no serious risk of bias	serious¹	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE¹ due to inconsistency	464/1281 (36.2%)	327/899 (36.4%)	OR 1.12 (0.93 to 1.35)	Study population
										362 per 1000	27 more per 1000 (from 17 fewer to 72 more)
										Moderate
											-
amenorrhoea after 2 years (assessed with: PBAC chart/Amenorrhoea rate) Batthacharya 2012
370 (2 studies) 2 years	no serious risk of bias	serious²	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE² due to inconsistency	75/187 (40.1%)	58/183 (31.7%)	OR 0.64 (0.41 to 0.99)	Study population
										401 per 1000	101 fewer per 1000 (from 2 fewer to 186 fewer)
										Moderate
											-
amenorrhoea after 5 years (assessed with: Menstrual status) Batthacharya 2012
236 (1 study) 1 years	serious³	no serious inconsistency	serious ⁴	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^3,4 due to risk of bias, indirectness	76/117 (65%)	82/119 (68.9%)	OR 1.19 (0.7 to 2.05)	Study population
										650 per 1000	38 more per 1000 (from 85 fewer to 142 more)
										Moderate
											-
Proportion with heavy bleeding after 1 year (assessed with: PBAC chart) Batthacharya 2012
2180 (13 studies) 1 years	no serious risk of bias	no serious inconsistency	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊕ HIGH	151/1281 (11.8%)	111/899 (12.3%)	OR 0.97 (0.74 to 1.28)	Study population
										118 per 1000	4 fewer per 1000 (from 31 fewer to 33 more)
										Moderate
											-
Proportion with heavy bleeding after 2 years (assessed with: PBAC chart) Batthacharya 2012
380 (2 studies) 2 years	no serious risk of bias	no serious inconsistency	no serious indirectness	no serious imprecision	serious⁵	⊕⊕⊕⊝ MODERATE⁵ due to publication bias	24/197 (12.2%)	33/183 (18%)	OR 0.54 (0.3 to 0.97)	Study population
										122 per 1000	52 fewer per 1000 (from 3 fewer to 82 fewer)
										Moderate
											-
Satisfaction after 1 year (assessed with: very/moderate/not satisfied) (Cochrane Lethaby 2009)
1690 (11 studies) 1 years	no serious risk of bias	no serious inconsistency	no serious indirectness	serious⁶	undetected	⊕⊕⊕⊝ MODERATE⁶ due to imprecision	904/990 (91.3%)	619/700 (88.4%)	OR 1.20 (0.85 to 1.7)	Study population
										913 per 1000	13 more per 1000 (from 14 fewer to 34 more)
										Moderate
											-

¹ I2 is 74%
² I2 is 36%
³ Inclusion and exclusion criteria not given
⁴ not given how menstrual status was defined
⁵ one study reported results after 3 years and not after 2 years

⁶not given how satisfaction was measured

Question: Should first/second generation ablation vs LNG-IUD be used for HMB? Bibliography: Bhattacharya 2011Hysterectomy, endometrial ablation and Mirena for heavy menstrual bleeding: a systematic review of clinical effectiveness and cost-effectiveness analysis
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
							With Mirena	With First/second generation ablation		Risk with Mirena	Risk difference with First/second generation ablation (95% CI)
amenorrhoea after 1 year (assessed with: PBAC chart/Amenorrhoea rate)
304 (6 studies) 1 years	serious¹	serious²	no serious indirectness	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^1,2 due to risk of bias, inconsistency	29/144 (20.1%)	32/160 (20%)	OR 1.02 (0.56 to 1.87)	Study population
										201 per 1000	3 more per 1000 (from 78 fewer to 119 more)
										Moderate
											-
amenorrhoea after 2 years (assessed with: PBAC chart/Amenorrhoea rate)
83 (2 studies) 2 years	serious¹	serious³	no serious indirectness	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^1,3 due to risk of bias, inconsistency	14/42 (33.3%)	9/41 (22%)	OR 1.79 (0.69 to 4.66)	Study population
										333 per 1000	139 more per 1000 (from 77 fewer to 366 more)
										Moderate
											-
Proportion with heavy bleeding after 1 year (assessed with: PBAC chart)
319 (6 studies) 1 years	serious¹	serious⁴	no serious indirectness	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^1,4 due to risk of bias, inconsistency	12/152 (7.9%)	17/167 (10.2%)	RR 1.36 (0.62 to 2.97)	Study population
										79 per 1000	28 more per 1000 (from 30 fewer to 156 more)
										Moderate
											-
QOL after 1 year (change from baseline) (measured with: SF 36 general health; range of scores: 0-100; Better indicated by higher values)
81 (2 studies) 1 years	serious¹	serious⁵	no serious indirectness	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^1,5 due to risk of bias, inconsistency	n.a.	n.a.	MD -0.08 (-5.17 to 5.02)		The mean QOL score after 1 year in the intervention groups was comparable to the control group
Re-interventions (hysterectomy) after 1 year (assessed with: proportion)
175 (3 studies) 1 years	serious¹	no serious inconsistency	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE¹ due to risk of bias	6/89 (6.7%)	2/86 (2.3%)	OR 0.36 (0.09 to 1.48)	Study population
										67 per 1000	42 fewer per 1000 (from 61 fewer to 29 more)
										Moderate
											-
Satisfaction (assessed with: satisfied yes/no)
265 (4 studies) 1 years	no serious risk of bias	serious⁶	serious⁷	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^6,7 due to inconsistency, indirectness	25/137 (18.2%)	22/128 (17.2%)	OR 0.94 (0.5 to 1.77)	Study population
										182 per 1000	9 fewer per 1000 (from 82 fewer to 101 more)
										Moderate
											-

¹ Half of studies randomisation not adequate
² I2 is 59%
³ I2 is 80%
⁴ I2 is 55%
⁵ I2 is 59%
⁶ I2 is 54%
⁷ indirect comparison

Question: Should embolisation vs surgery be used for HMB?
Bibliography: van der Kooij 2011 (EMMY trial) and Moss 2011 (REST trial)

Quality assessment

Summary of Findings

Participants
(studies)
Follow up

Risk of bias

Inconsistency

Indirectness

Imprecision

Publication bias

Overall quality of evidence

Study event rates (%)

Relative effect
(95% CI)

Anticipated absolute effects

With Surgery

With Embolisation

Risk with Surgery

Risk difference with Embolisation (95% CI)

QOL score after 5 years (measured with: SF 36; range of scores: 0-100; Better indicated by higher values)

282
(2 studies)
5 years

no serious risk of bias

no serious inconsistency

no serious indirectness

no serious imprecision

undetected

⊕⊕⊕⊕
HIGH

n.a.

MD 1.51

(0.58 to 3.6)

The mean QOL score after 5 years in the intervention groups was comparable to the control group

Re-interventions (assessed with: questionnaire)

313
(2 studies)
5 years

no serious risk of bias

serious¹

no serious indirectness

no serious imprecision

undetected

⊕⊕⊕⊝
MODERATE¹
due to inconsistency

9/126
(7.1%)

56/187
(29.9%)

OR 6.51
(2.98 to 14.24)

Study population

71 per 1000

262 more per 1000
(from 115 more to 451 more)

Moderate

Hospital stay (UAE versus hysterectomy) (measured with: days; range of scores: 0-7; Better indicated by lower values)

266
(3 studies)
2 years

no serious risk of bias

serious²

no serious indirectness

no serious imprecision

undetected

⊕⊕⊕⊝
MODERATE²
due to inconsistency

n.a.

-2.83
(-3.21 to -2.45)

The mean hospital stay (uae versus hysterectomy) in the intervention groups was
2.83 lower
(3.21 to 2.45 lower)

Hospital stay (UAE versus myomectomy) (measured with: days; range of scores: 1-4; Better indicated by lower values)

121
(1 study)
6 months

serious³

no serious inconsistency

no serious indirectness

serious⁴

undetected

⊕⊕⊝⊝
LOW^3,4
due to risk of bias, imprecision

n.a.

-1.10
(-1.64 to -0.56)

The mean hospital stay (uae versus myomectomy) in the intervention groups was
1.10 lower
(1.64 to 0.56 lower)

¹ I2 is 53%
² I2 is 79%
³ randomization method not well described
⁴ outcome measures not well defined

Question: Should laparoscopic uterine artery occlusion vs embolisation be used for myomas? Bibliography: Ambat 2009 Uterine artery embolization versus laparoscopic occlusion of uterine vessels for management of symptomatic uterine fibroids
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
							With Embolisation	With Laparoscopic uterine artery occlusion		Risk with Embolisation	Risk difference with Laparoscopic uterine artery occlusion (95% CI)
Ruduction bloodloss (assessed with: PBAC reduction in %)
56 (1 study) 6 months	serious risk of bias¹	no serious inconsistency	no serious indirectness	very serious²	undetected	⊕⊕⊝⊝ LOW^1,2 due to risk of bias, imprecision	50% reduction	56% reduction	Not given	Study population
										See comment	-
										Moderate
											-

¹ small study group
² no confidence intervals given

Question: Should ablation with GnRH analogue pretreatment vs ablation alone be used for HMB? Bibliography: CochraneSowter 2009 Pre-operative endometrial thinning agents before endometrial destruction for heavy menstrual bleeding
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
							With Ablation alone	With Ablation with GnRH analogue pretreatment		Risk with Ablation alone	Risk difference with Ablation with GnRH analogue pretreatment (95% CI)
amenorrhoea after 1 year (assessed with: PBAC chart/Amenorrhoea rate)
605 (7 studies) 1 years	serious¹	serious²	no serious indirectness	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^1,2 due to risk of bias, inconsistency	76/309 (24.6%)	115/296 (38.9%)	RR 1.58 (1.19 to 2.01)	Study population
										246 per 1000	143 more per 1000 (from 47 more to 248 more)
										Moderate
											-
amenorrhoea after 2 years (assessed with: PBAC chart/Amenorrhoea rate)
357 (2 studies) 2 years	very serious^1,2	no serious inconsistency	no serious indirectness	no serious imprecision	undetected	⊕⊕⊝⊝ LOW¹ due to risk of bias	26/181 (14.4%)	41/176 (23.3%)	RR 1.62 (1.04 to 2.52)	Study population
										144 per 1000	89 more per 1000 (from 6 more to 218 more)
										Moderate
											-
Satisfaction (assessed with: very/moderate/not satisfied)
599 (6 studies)	Serious³	no serious inconsistency	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE¹ due to risk of bias	265/301 (88%)	260/298 (87.2%)	RR 0.99 (0.93 to 1.05)	Study population
										880 per 1000	9 fewer per 1000 (from 62 fewer to 44 more)
										Moderate
											-

¹ Half of studies unclear allocation concealment

²Small study group
³ I2 is 40%

NB: in the plots the favours (control and GnRH) were turned around

Question: Should 5 mg ulipristal vs placebo be used for HMB (myomas)? Bibliography: Donnez 2012 Ulipristal Acetate versus Placebo for Fibroid Treatment before Surgery
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
							With Placebo	With 5 mg ulipristal		Risk with Placebo	Risk difference with 5 mg ulipristal (95% CI)
PBAC < 75 (measured with: PBAC chart; range of scores: 0-500; Better indicated by lower values, % patients with PBAC <75)
143 (1 study) 13 weeks	serious¹	no serious inconsistency	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE¹ due to risk of bias	9/48 (19%)	86/94 (91%)	n.a.	9 of 48 patients PBAC < 75 in the control group	in 73 patients the mean PBAC was below 75 in the intervention groups was (55 to 83 patients had PBAC< 75)
Change fibroid volume (measured with: MRI ; range of scores: 0-100; Better indicated by higher value, % volume )
143 (1 study) 13 weeks	serious¹	no serious inconsistency	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE¹ due to risk of bias	3.0%	-21%	n.a.	The mean change fibroid volume in the control groups was 3 % change	The mean change fibroid volume in the intervention groups was 22.6 % lower (36.1 to 8.2 lower)

¹ Small study group

Question: Should ulipristal vs GnRH analogue be used for HMB (myomas)? Bibliography: Donnez 2012 Ulipristal acetate versus leuprolide acetate for uterine fibroids.
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
							With GnRH analogue	With Ulipristal		Risk with GnRH analogue	Risk difference with Ulipristal (95% CI)
PBAC < 75 (assessed with: PBAC pictorial chart)
185 (1 study)	serious¹	no serious inconsistency	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE¹ due to risk of bias	82/92 (89.1%)	84/93 (90.3%)	Dfference 1.2% (-9.3 to 11.8)	Study population
										82 of 92 patients PBAC < 75 in the control group	1.2 (-9.3 to 11.8)


Change uterine volume (measured with: MRI; Better indicated by lower values)
186 (1 study)	serious¹	no serious inconsistency	no serious indirectness	no serious imprecision	undetected	⊕⊕⊕⊝ MODERATE¹ due to risk of bias	-47%	-20%	Difference Not given	The mean change uterine volume in the control groups was 47 %	The mean change uterine volume in the intervention groups was 20%

¹ small study Group

Question: Should GnRH analogues before TCRM vs placebo be used for submucous myomas? Bibliography: Mavrelos 2010 The value of pre-operative treatment with GnRH analogues in women with submucous fibroids: a double-blind, placebo-controlled randomized trial
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
							With Placebo	With GnRH analogues before TCRM		Risk with Placebo	Risk difference with GnRH analogues before TCRM (95% CI)
complete resection (assessed with: hysteroscopic view)
47 (1 study) 6 weeks	serious¹	no serious inconsistency	no serious indirectness	serious²	undetected	⊕⊕⊝⊝ LOW^1,2 due to risk of bias, imprecision	16/23 (69.6%)	14/24 (58.3%)	RR 0.84 (0.54 to 1.29)	Study population
										696 per 1000	111 fewer per 1000 (from 320 fewer to 202 more)
										Moderate
											-

¹ Outcome measure not well defined
² Few patients and few events

Question: Should GnRH analogues vs placebo/no treatment be used before myomectomy? Bibliography: Lethaby 2011. Pre-operative GnRH analogue therapy before hysterectomy or myomectomy for uterine fibroids (Review)
Quality assessment							Summary of Findings
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	Study event rates (%)		Relative effect (95% CI)	Anticipated absolute effects
Participants (studies) Follow up	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall quality of evidence	With Placebo/no treatment	With GnRH analogues	Relative effect (95% CI)	Risk with Placebo/no treatment	Risk difference with GnRH analogues (95% CI)
Change fibroid volume (measured with: ultrasound ; range of scores: 0-250; Better indicated by higher values)
461 (6 studies) 3 months	serious¹	serious²	no serious indirectness	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^1,2 due to risk of bias, inconsistency	n.a.	n.a.	MD 12.49	The mean change fibroid volume in the control groups was 164.8 mililitres	The mean change fibroid volume in the intervention groups was 12.49 mililitres lower (18.34 to 6.64 lower)
Intra-operative bloodloss (measured with: mililiters; range of scores: 0-300; Better indicated by lower values)
179 (4 studies) 3 months	no serious risk of bias	serious³	serious⁴	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^3,4 due to inconsistency, indirectness	n.a.	n.a.	MD 67.5	The mean intra-operative bloodloss in the control groups was 341.4 mililiters	The mean intra-operative bloodloss in the intervention groups was 67.5 mililitres lower (18.43 to 44.4 lower)
Duration of surgery (measured with: minutes; range of scores: 0-200; Better indicated by lower values)
190 (5 studies) 3 months	no serious risk of bias	serious⁵	serious⁴	no serious imprecision	undetected	⊕⊕⊝⊝ LOW^4,5 due to inconsistency, indirectness	n.a.	n.a.	MD 4.20	The mean duration of surgery in the control groups was 100.1 minutes	The mean duration of surgery in the intervention groups was 4.20 higher (2.69 lower to 11.8 higher)

¹ In almost all trials randomisation not adequate; ² I2 87%; ³ I2 is 39%; ⁴ not always clear when written about myomectomy of hysterectomy; ⁵ I2 is 70%