Participants with observations from at least one of these
timepoints were included in the analyses. Baseline
EQ-5D scores and intraocular pressure values were used
as explanatory variables. The model included fi xed eff ects
for sex, ethnic origin, diagnosis, whether glaucoma was
present in one or both eyes, and intervention. Dummy
variables for the timepoint were included to enable
investigation of the eff ects of the interventions at each
timepoint. Random eff ects were included for centre and
individual. The model was extended for subgroup
analyses by fi tting a dummy variable for each respective
subgroup. These dummy variables were used to create
further interaction terms to represent the eff ect of clear-
lens extraction in the subgroups at each of the timepoints,
expressed as odds ratios and 95% CIs.
The secondary continuous and binary outcomes were
analysed with appropriate generalised linear models.
The unit of analysis for the clinical outcomes was the
treated eye (the worse eye if both were suitable for
treatment). For quality of life measures, the unit of
analysis was the participant, with bilateral disease
included as a fi xed eff ect covariate. To account for
missing answers in questionnaires we followed the
authors’ recommendations. These allow a score to be
generated if there are missing questions in the
NEI-VFQ-25, whereas for EQ-5D and the glaucoma-
specifi c disability questionnaire, no score is assigned.
Planned subgroup analyses used the minimisation
variables ethnic origin (Chinese or non-Chinese),
diagnosis (primary angle closure or primary angle-closure
glaucoma), and unilateral or bilateral disease. We added
an unplanned subgroup analysis after baseline visual
acuity data were assessed to explore the possible diff erence
in the primary outcome between patients with excellent
and slightly decreased visual acuity (≥85 ETDRS letters vs
<85 ETDRS letters).
The in-trial cost-eff ectiveness data were obtained with
seemingly unrelated regression adjustment for baseline
cost and EQ-5D score. We compared mean costs and
eff ects to estimate the incremental cost-eff ectiveness ratio
(ICER) for clear-lens extraction versus standard care.
85>
Chia sẻ với bạn bè của bạn: |