Table 3.
Effect Estimates from Univariable and Multivariable Models of Distance Visual Acuity in Study Eye during Study Eye Monitoring
| Variable | Univariable |
Multivariable |
||||
|---|---|---|---|---|---|---|
| MD (95% CI) | P Value | MD (95% CI) | P Value | P Value for Interaction with Time | ||
| Time (per year), multivariable model | −4.3 (−4.9 to −3.7) | - | - | |||
| Time (per year) | −4.3 (−4.9 to −3.7) | - | ||||
| Age at IVAN exit (per 10 yrs), centered | −3.8 (−5.9 to −1.7) | - | −1.1 (−2.1 to −0.1) | - | - | |
| Age at IVAN exit (per 10 yrs), centered × time | −1.2 (−2.1 to −0.4) | 0.005 | −1.3 (−2.1 to −0.5)‡ | - | 0.001 | |
| Time (per year) | −4.3 (−5.1 to −3.5) | - | ||||
| Gender (male) | 1.0 (−2.2 to 4.2) | - | 0.5 (−1.0 to 1.9) | 0.529 | 0.916 | |
| Gender (male) × time | 0.4 (−0.9 to 1.7) | 0.579 | ||||
| Time (per year) | −2.8 (−4.1 to −1.5) | - | ||||
| Index of multiple deprivation decile | 0.4 (−0.2 to 0.9) | - | −0.2 (−0.4 to 0.1) | 0.193 | 0.100§ | |
| Index of multiple deprivation decile × time | −0.2 (−0.4 to 0.0) | 0.017 | ||||
| Time (per year) | −4.2 (−4.9 to −3.4) | - | ||||
| BCVA at IVAN exit | ≥68 | Ref. | - | Ref. | <0.001 | 0.398 |
| 53–67 | −18.2 (−20.2 to −16.2) | −17.0 (−19.0 to −15.1) | ||||
| 38–52 | −31.3 (−33.9 to −28.7) | −29.5 (−32.0 to −27.0) | ||||
| ≤37 | −46.5 (−48.9 to −44.0) | −44.6 (−47.0 to −42.1) | ||||
| BCVA at IVAN exit × time | ≥68 | Ref. | 0.326 | |||
| 53−67 | −0.9 (−2.7 to 0.8) | |||||
| 38−52 | 1.4 (−0.8 to 3.6) | |||||
| ≤37 | 0.4 (−1.9 to 2.8) | |||||
| Time (per year) | −4.5 (−5.4 to −3.6) | - | ||||
| nAMD present in fellow eye | 3.4 (0.2 to 6.5) | - | −0.9 (−2.6 to 0.7) | 0.281 | 0.202 | |
| nAMD present in fellow eye × time | 0.6 (−0.6 to 1.9) | 0.324 | ||||
| Time (per year) | −4.4 (−5.2 to −3.6) | - | ||||
| Study eye BCVA better than fellow eye at IVAN exit∗ | 12.6 (9.5 to 15.7) | - | 3.9 (2.0 to 5.7) | <0.001 | 0.172 | |
| Study eye BCVA better than fellow eye at IVAN exit × time | 0.6 (−0.7 to 1.9) | 0.371 | ||||
| Time (per year) | −4.2 (−4.8 to −3.6) | - | ||||
| Injection rate in study eye in previous year (per 3 injections) | −0.1 (−0.8 to 0.7) | - | −0.2 (−0.9 to 0.5) | 0.549 | 0.056‖ | |
| Injection rate in study eye in previous year (per 3 injections) × time | 0.2 (0.0 to 0.3) | 0.056 | ||||
| Time (per year) | −4.6 (−5.3 to −3.8) | - | ||||
| Proportion change in lesion size† | 0.1 (−0.4 to 0.5) | - | 0.0 (−0.1 to 0.1) | 0.757 | 0.178 | |
| Proportion change in lesion size × time | 0.1 (−0.1 to 0.4) | 0.159 | ||||
BCVA = best-corrected visual acuity; CI = confidence interval; DVA = distance visual acuity; IMD = index of multiple deprivation; IVAN = Inhibition of VEGF in Age-related choroidal Neovascularisation; MD = mean difference; nAMD = neovascular age-related macular degeneration.
Injection rate in study eye in previous year (per 3 injections) × time = 0.2 (0.0–0.3)
Study eye is defined as better than the fellow eye if study eye BCVA ≥5 letters greater than fellow eye BCVA at IVAN exit.
Proportion change in lesion size between IVAN entry and IVAN exit (lesion size at IVAN exit/lesion size at IVAN entry).
This interaction means that in an 80-year-old, the average rate of change in DVA in the study eye was −4.3 (95% CI, −4.9 to −3.7) letters per year but, for every 10-year increase in age, a further 1.3 letters per year were lost; thus, for a 90-year-old, the rate of change in DVA was −4.3 to 1.3 = −5.6 letters per year. See also Figure 4.
Interaction not included in final multivariable model. Effect estimates for IMD (main effect) and IMD by time interaction if included in the multivariable model are index of multiple deprivation decile = −0.1 (−0.4 to 0.1). Index of multiple deprivation decile × time = −0.2 (−0.4 to 0.1). Model fitted to n = 532 patients. Missing data imputed using multiple imputation methods. Table S10 gives complete case analysis. Age is centered at the average age (80 years). Because of small effect sizes for each 1-year increase in age, age is scaled per 10 years. Because of the small effect size for each increase in 1 injection, injection rate is scaled, and effect sizes reported are per 3 injections. Univariable models were fitted with time, covariate and time × covariate interaction only. Multivariable models were fitted with time, all covariates and time × age interaction.
Interaction not included in final multivariable model. Effect estimates for injection rate (main effect) and injection rate by time interaction if included in the final multivariable model are injection rate in study eye in previous year (per 3 injections) = −0.5 (−1.2 to 0.3).