Table 3.

Evaluation of psychometric characteristics of the patient-reported outcome measures in presbyopia

PROMs	Validity: content		Item-level performance using Rasch analysis	Reliability		Validity: construct			Responsiveness in presbyopia¶	Comments
	Patient input	Cognitive debriefing		Internal consistency*	Test–retest†	Hypothesis specified	Convergent/divergent‡	Known groups§
NAVQ (2008)	√	NR	√	√	√	√	√	√	√	Only measure found to be specific to presbyopia. Has rigorous and well-documented development and has robust psychometric properties.
NEI VFQ-9 (2010)	√	NR	√	√	NR	√	√	√	NR	Demonstrates stronger psychometric properties than the original 25-item version.
NEI VFQ-25 (2001)	√	NR	X	X	X	√	√	√	X	Failed to demonstrate validity as it did not fit the Rasch model; several subscales found to be psychometrically unsound. Lengthy and not easy to complete.
QoV (2010)	√	√	√	√	√	NR	√	√	√	Although this measure has strong psychometric properties, the validation study suffered from methodological limitations, such as the average value of the measurements for both eyes was used for correlation analysis for construct validity assessment.
VF-14 (1994)	NR	NR	√	NR	NR	NR	NR	NR	√	Originally designed for measuring functional impairments caused by only cataracts. Not validated in individuals with presbyopia. Has a two-stage question structure, which is likely to make it more confusing to complete.
Cataract TyPE Spec (2003)	X	NR	X	NR	NR	NR	NR	NR	√	Developed to measure cataract surgery outcome. Not validated in a presbyopia sample. Significant floor effect in that the items did not target participants with low levels of visual disability (ie, at bottom of the item-person map).
Catquest-9SF (2008)	X	NR	√	NR	NR	NR	NR	NR	√	Not developed using input from individuals with presbyopia. Small level of differential item functioning was seen for a few items.
NEI RQL-42 (2003)	√	NR	X	√	√	X	NR	NR	X	Was shown to be deficient for all psychometric properties tested using Rasch analysis, with issues related to lack of precision in scores, item misfit and poor targeting.
FGVS (2010)	√	√	X	√	NR	√	√	√	NR	A ceiling effect was observed for the dimension as well as for all subdimension scores, that is, majority of individuals selecting highest response option for many of the items.
RSVP (2000)	√	NR	X	√	NR	NR	√	√	NR	Subscales suffer from poor targeting. 1 of 3 items in each scale demonstrates differential item functioning. This is a problem because it indicates that people of different characteristics answer questions differently.

√: meets the criteria for good psychometric performance; X: fails to satisfy the criteria for good psychometric performance.

*Range for acceptable Cronbach’s alpha: above 0.70 but not higher than 0.95.⁴⁴

†Threshold for acceptable test–retest reliability: interclass correlation coefficient of 0.75 or greater.⁴⁵

‡At least one Pearson’s correlation coefficient (r) value was categorised as moderate (0.25–0.50) or strong (>0.50).⁴⁶

§Discriminant validity demonstrated by statistically significant (p<0.05) difference in at least one comparison of patient subgroups with differing clinical features.

¶Responsiveness demonstrated by statistically significant (p<0.05) changes in scores following an intervention in at least one longitudinal validation study or one randomised controlled trial in individuals with presbyopia.

FGVS, Freedom From Glasses Value Scale; NAVQ, Near Activity Visual Questionnaire; NEI RQL-42, National Eye Institute Refractive Error Quality Of Life Instrument-42; NEI VFQ-9, National Eye Institute Visual Function Questionnaire-9; NEI VFQ-25, National Eye Institute Visual Function Questionnaire-25; NR, not reported; PROMs, patient-reported outcome measures; QoV, Quality of Vision; RSVP, Refractive Status And Vision Profile; VF-14, Visual Function Index.