Page 178 - DISINVESTMENT AND IMPLEMENTATION OF VISION SCREENING TESTS BASED ON THEIR EFFECTIVENESS

CHAPTER 9
and in one third more than two times. The large diversity in screening programmes is caused by the lack of cost-effectiveness studies, lack of data collection and quality monitoring due to the development of practise bases screening protocols separate for all countries and to the lack of competion in preventive healthcare. Comparison of vision screening programmes is hampered by insufficient data collection and monitoring, that impedes comparison and again perpetuates their diversity.
The Dutch vision screening programme is very extensive with seven exams. Comparison and evaluation of existing screening programmes is difficult, especially for programmes with repeated screens. When only one measurement of VA is performed the specificity, sensitivity and attendance can be estimated and the cost-effectiveness calculated, but with repeated screens this calculation is difficult. A established method to compare the cost-effectiveness of screening programmes is the use of micro-simulation models (MISCAN). With a micro-simulation model, the effectiveness of introduction, modification or disinvestment of (components of) a screening programme can be calculated, provided that very detailed data is available. We used the detailed data from the RAMSES birth cohort study for a micro-simulation model. In the calculation with the model the uncertainty of the age of onset of amblyopia proved to be important. As the incidence curve of amblyopia is unknown, we made an estimation based on approximation of the observational data from the RAMSES study in conjunction with experts’ estimations and literature. For each type of amblyopia, deprivation, strabismus, refractive error and combined strabismus and refractive error, the mean actual sensitivity was estimated. The preverbal orthoptic vision screens had lower sensitivity than the VA measurements. Using the incidence curves and the mean sensitivity per screen, we estimated the effect per screen. The finding in the RAMSES birth cohort study that orthoptic tests at age 6-24 months yielded very few cases of amblyopia was confirmed by the micro-simulation model. The sensitivity and specificity of the entire screening programme was not appreciably affected by the omission of the preverbal vision screens. The observations in the RAMSES study, that orthoptic vision screening tests contributed little to the detection of refractive amblyopia, while strabismus amblyopia was detected by the parents, was confirmed by the results of the model simulation.
The disappointing yield of the preverbal vision screening test was reason for on-site evaluation of all vision screening tests in the child healthcare (CHC) setting. Three sources were used for this evaluation, the screening records, questionnaires and semi-structured observations of all vision screening tests, as defined in the national
176