The guidelines manual

Quality assessment

No. of patients

Effect

Quality

Importance

No. of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

Duloxetine

Placebo

Relative
(95% CI)

Absolute

Patient-reported 30% pain reduction (follow-up 12 weeks)

2¹

Randomised trials

No serious risk of bias

Serious²

No serious indirectness

No serious imprecision

None

220/327

111/215

RR 1.33
(0.95 to 1.88)

17 more per 100 (from 3 fewer to 45 more)

Moderate

Critical

Patient-reported 50% pain reduction (follow-up 12 weeks)

4³

Randomised trials

No serious risk of bias

Serious⁴

No serious indirectness

Serious imprecision⁵

None

485/896

164/443

RR 1.51 (1.17 to 1.94)

19 more per 100 (from 6 more to 35 more)

Moderate

Critical

No. of withdrawals due to adverse effects (follow-up 12 weeks)

4³

Randomised trials

No serious risk of bias

No serious inconsistency

No serious indirectness

No serious imprecision

None

113/906

21/448

RR 2.63 (1.68 to 4.12)

8 more per 100 (from 3 more to 15 more)

High

Critical

Dizziness (adverse effects) (follow-up 12 weeks)

3⁶

Randomised trials

No serious risk of bias

No serious inconsistency

No serious indirectness

Serious imprecision⁵

None

90/674

26/332

RR 1.81 (1.17 to 2.79)

6 more per 100 (from 1 more to 14 more)

Moderate

Critical

Dry mouth (adverse effects) (follow-up 12 weeks)

2⁷

Randomised trials

No serious risk of bias

No serious inconsistency

No serious indirectness

Serious imprecision⁵

None

37/448

10/224

RR 1.61 (0.82 to 3.20)

3 more per 100 (from 1 fewer to 10 more)

Moderate

Important

GI disturbances (adverse effects) (follow-up 12 weeks)

2⁸

Randomised trials

No serious risk of bias

No serious inconsistency

No serious indirectness

Serious imprecision⁵

None

28/332

8/217

RR 2.53 (1.13 to 5.67)

6 more per 100 (from 0 more to 17 more)

Moderate

Important

Any adverse effects (non-specified) (follow-up 12 weeks)

1⁹

Randomised trials

No serious risk of bias

No serious inconsistency

No serious indirectness

Very serious imprecision¹⁰

None

86/106

78/109

RR 1.13 (0.98 to 1.32)

9 more per 100 (from 1 fewer to 23 more)

Low

Critical

¹ Gao et al. (2010); Wernicke et al. (2006).

² Substantial heterogeneity, random-effect model was used. Potential sources of heterogeneity: i) Gao et al. (2010) – ITT data available, used flexible dose between 30 mg and 120 mg, non-pharmaceutical company funded; ii) Wernicke et al. (2006) – only per-protocol data available, combined two fixed doses (60 mg and 120 mg), pharmaceutical company funded.

³ Gao et al. (2010); Goldstein et al. (2005); Raskin et al. (2005); Wernicke et al. (2006).

⁴ Substantial heterogeneity, random-effect model was used. Potential sources of heterogeneity: i) Gao et al. (2010) – used flexible dose between 30 mg and 120 mg, non-pharmaceutical company funded; ii) Goldstein et al. (2005), Raskin et al. (2005) and Wernicke et al. (2006) – combined different fixed doses (20 mg, 60 mg and 120 mg), pharmaceutical company funded.

⁵ Confidence interval crossed one end of default MID.

⁶Gao et al. (2010); Goldstein et al. (2005); Wernicke et al. (2006).

⁷ Gao et al. (2010); Goldstein et al. (2005).

⁸ Gao et al. (2010); Wernicke et al. (2006).

⁹ Gao et al. (2010).

¹⁰ Confidence interval crossed both ends of default MID.

Abbreviations: CI, confidence interval; GI, gastrointestinal; ITT, intention to treat; MID, minimal important difference; RR, relative risk.

Quality assessment

No. of patients

Effect

Quality

Importance

No of studies

Design

Risk of bias

Inconsistency

Indirectness

Imprecision

Other considerations

.

.

Relative (95% CI)

Absolute

X

X

X

X

X

X

[References, abbreviations and other footnotes].

Study

Limitations

Applicability

Other comments

Incremental

Uncertainty

Costs

Effects

ICER

Noble 1998 Budesonide CIR versus no maintenance therapy

Potentially serious limitations^1,2

Partially applicable³

Study employed a Markov decision-analytic model with a 1-year time horizon

£115

0.017 QALYs⁵

£6,981 per QALY gained

ICER decreases significantly if the cost of surgery is increased.

NCGC model

Oral budesonide versus no maintenance therapy⁴

Potentially serious limitations²

Directly applicable

Study employed a Markov decision-analytic model with a 2-year time horizon

£477^f

£150⁷

£528⁸

£336⁹

0.012 QALYs⁶

0.012 QALYs⁷

0.006 QALYs⁸

0.005 QALYs⁹

£40,392 per QALY gained⁶

£15,070 per QALY gained⁷

£87,610 per QALY gained⁸

£65,013 per QALY gained⁹

No treatment most cost-effective option when baseline risk of relapse decreased.

In the PSA, probability of budesonide being the most cost-effective treatment at willingness-to-pay threshold of £20,000 per QALY gained ranged from 0 to 8%

¹ Modelling was undertaken over a short time horizon and no probabilistic sensitivity analysis was conducted.

² Specific costs and disutilities of drug-related adverse events could not be explicitly modelled. Adverse events were captured by modelling treatment-specific withdrawal rates. This may have overestimated the cost effectiveness of maintenance treatment.

³ The cost-effectiveness model was designed to reflect the management of Crohn's disease in the Swedish healthcare setting. Although a cost per QALY estimate was reported, it was not based on health-related quality of life values elicited from patients.

⁴ The NCGC model compared a number of different maintenance treatments.

⁵ Figures may differ because of rounding off.

⁶ Conservative 4-line model. Conservative treatment effects were used and people relapsing while on azathioprine maintenance treatment had a different induction sequence.

⁷ Conservative 3-line model. Conservative treatment effects were used and people were assumed to have the same induction sequence regardless of maintenance treatment.

⁸ Non-conservative 4-line model. Non-conservative treatment effects were used and people relapsing while on azathioprine maintenance treatment had a different induction sequence.

⁹ Non-conservative 3-line model. Conservative treatment effects were used and people were assumed to have the same induction sequence regardless of maintenance treatment.

Study

Limitations

Applicability

Other comments

Incremental

Uncertainty

Costs

Effects

ICER

.

.

.

.

.

[References, abbreviations and other footnotes].