Table 1 shows the baseline characteristics of the patients in the present study. Among 75 eyes in 75 CSC patients (58 males and 17 females) with serous retinal detachment (SRD), 45 patients were smokers and 30 patients were non-smokers. Thirty-one patients were current smokers, and the remaining 44 patients were non-current smokers. The mean patient age was 47.1 ± 6.0 years (mean ± standard deviation). No significant difference in age was seen between the smoking group (47.2 ± 6.3 years) and the non-smoking group (46.8 ± 5.5 years, p = 0.74, Wilcoxon rank sum test). The Brinkman index (BI, number of cigarettes smoked per day multiplied by the number of years of smoking) was 221.9 ± 238.6 in the smokers group. The mean spherical equivalent of the refractive error also showed no significant difference between the smoking and non-smoking groups (-1.31 ± 1.80 diopter vs. -2.06 ± 1.84 diopter, p = 0.10). A significant difference in the SFCT was seen between the smoker and non-smoker groups (382.0 ± 68.2 µm vs. 339.3 ± 52.3 µm, p = 0.0038, Wilcoxon rank sum test, Fig. 1A), while no significant difference was observed in the CVI (64.6 ± 2.1% vs. 65.3 ± 2.3%, p = 0.32, Fig. 1B).
Table 1
Baseline characteristics of patients
Parameter | Smoking group | Non-smoking group | P value |
Number of patients, eyes | 45, 45 | 30, 30 | - |
Age (years) BI | 47.2 ± 6.3 221.9 ± 238.6 | 46.8 ± 5.5 - | 0.74 - |
SE (diopter) | -1.31 ± 1.80 | -2.06 ± 1.84 | 0.10 |
LogMAR VA | 0.057 ± 0.24 | -0.000089 ± 0.21 | 0.11 |
SFCT (µm) | 382.0 ± 68.2 | 339.3 ± 52.3 | 0.0038 |
CVI (%) | 64.6 ± 2.1 | 65.3 ± 2.3 | 0.32 |
FRT (µm) | 367.8 ± 106.5 | 360.3 ± 130.1 | 0.69 |
BI: Brinkman index, SE: spherical equivalent, logMAR VA: logarithm of the minimum angle of resolution of visual acuity, SFCT: subfoveal choroidal thickness, FRT: foveal retinal thickness, CVI: choroidal vascular index. |
We then investigated factors associated with SFCT. As a result of the bias-corrected Akaike’s information criterion (AICc) model selection, the optimal model for SFCT included the variables of age, refractive error expressed by the spherical equivalent (SE), and history of smoking among the variables of age, SE, and history of smoking or hypertension (Table 2). The optimal model formula was as follows:
Table 2
Relationship between smoking and SFCT
| Univariate analysis | The optimal model |
Variables | Coefficient | Stderr | P value | Coefficient | Stderr | P value |
Age | -3.39 | 1.22 | 0.0070 | -3.40 | 1.12 | 0.0034 |
SE | 10.64 | 3.97 | 0.0091 | 8.27 | 3.71 | 0.029 |
History of smoking | 42.7 | 14.7 | 0.0049 | 37.7 | 13.9 | 0.0082 |
History of hypertension | 8.43 | 22.36 | 0.71 | N.S. | N.S. | N.S. |
SFCT: subfoveal choroidal thickness, Stderr: standard error, SE: spherical equivalent, N.S.: not selected. |
SFCT = 515.7–3.40 x Age (standard error [Stderr] = 1.12, p = 0.0034) + 8.27 x SE (Stderr = 3.71, p = 0.029) + 37.7 x Smoking (Stderr = 13.9, p = 0.0082) (AICc = 827.6).
In the aforementioned analysis, a history of smoking, but not BI, was included in the variable due to colinearity between these two factors. When this analysis was performed using the BI instead of a history of smoking (Table 3), BI was selected as a predictive variable, and the formula for the optimal model for SFCT was as follows:
Table 3
Relationship between BI and SFCT
| Univariate analysis | The optimal model |
Variables | Coefficient | Stderr | P value | Coefficient | Stderr | P value |
Age | -3.39 | 1.22 | 0.0070 | -4.47 | 1.19 | 0.00034 |
SE | 10.64 | 3.97 | 0.0091 | 7.79 | 3.72 | 0.040 |
BI | 0.063 | 0.031 | 0.046 | 0.088 | 0.030 | 0.0052 |
History of hypertension | 8.43 | 22.36 | 0.71 | N.S. | N.S. | N.S. |
BI: Brinkman index, SFCT: subfoveal choroidal thickness, Stderr: standard error, SE: spherical equivalent, N.S.: not selected. |
SFCT = 568.5–4.47 x Age (Stderr = 1.19, p = 0.00034) + 7.79 x SE (Stderr = 3.72, p = 0.040) + 0.088 x BI (Stderr = 0.030, p = 0.0052) (AICc = 826.7).
On the other hand, the optimal model for CVI included SE and history of smoking among the baseline parameters. The formula for the optimal model was as follows:
CVI = 65.9 + 0.29 x SE (Stderr = 0.14, p = 0.039) – 0.90 x Smoking (Stderr = 0.51, p = 0.083) (AICc = 331.7)
However, unlike the results for SFCT, when the BI was used instead of a history of smoking, BI was not selected as an explanatory variable for CVI.
Moreover, we investigated the relationship between current smoking and the choroidal structure. No significant differences in SFCT and CVI were seen between current smokers and non-current smokers (p = 0.56 [Figure 2] and 0.98, respectively, Wilcoxon rank sum test). Furthermore, current smoking was not selected for the SFCT and the CVI using AICc model selection (data not shown).