5
434 AM J OPTOM & PHYSIOL OPTICS Vol. 64, No. 6
FIG. 3. Mean asthenopia scores are presented on the ordinate, whereas phases of testing are presented on the abscissa. Open squares j’” represent patients who received experimental therapy first; closed squares represent those patients who received placebo therapy first.
TABLE 2. Asthenopia change scores after control or
experimental phases.
Phase 1
Training Phase
Patient No.
a
Group
Control Experimental b
1
Control
3
5
2
Experimental 0
8
3
Control
1
3
4
Control
—2
2
5
Experimental —1
5
Mean change
0.2
4.6
a.b See explanations given in Table 1. Scores in Table
2 represent asthenopia changes.
ment of accommodative amplitude and facility. It also indicates that asthenopia associated with accommodative anomalies can be reduced with such training. Neither the improvement in accommodative ability nor the decrease in asthenopia can be accounted for by placebo effects or experimental bias.
Previous studies have reported an 85% success rate with patients in the reduction of asthenopia after some form of accommodative training.12-15 In our study, four of five patients also experienced a significant relief of symptoms. The one patient who did not show a decrease in symptoms also did not show much increase in accommodative amplitude or facility after training. This finding further supports the notion that
improvement in accommodative amplitude and facility is related functionally to symptom reduction.
The major changes in performance associated with accommodative training were a reduction in reported blur during reading and an increase in reading time. The asthenopia questionnaire revealed a reduction in ocular fatigue, although results were not as dramatic. Overall, the patients in this study changed from moderately uncomfortable to reasonably comfortable in a relatively short period of time.
Our primary purpose was to determine if accommodative therapy reduces asthenopia while controlling for any source of experimental bias by using a placebo control paradigm. Although a small subject sample was used, that in itself should not limit the reliability or the validity of the findings. The matched-subjects crossover control design is powerful in controlling for sources of experimental confounding and biases.18’19 Not only are experimental and control treatments compared during the same pe riod of time, but the control group “crosses over” to receive the experimental treatment at a later time. This replication further supports the general conclusion of the study. Appropriate statistical analyses also show that findings are unlikely to occur by chance.
This study was not designed to determine the best way to treat an accommodative anomaly. Only a single therapeutic technique was used for

5