2.2. Patient Selection
Patients diagnosed with the following DVDs were enrolled in the study group of the current study: (1) Kaposi’s sarcoma (ICD-9 code: 176.x), (2) hemangiomas (ICD-9 code: 228.0x), (3) pyogenic granuloma (ICD-9 code: 686.1), and (4) teleangiectasias and superficial capillary disorders (ICD-9 code: 448.9). To further enhance the diagnostic accuracy in the study group, only patients received the above ICD-9 code from a dermatologist (department code: 11) were included to ensure the vascular lesions exactly as they came from the skin. In addition, patients were excluded if any of the criteria occurred to erase the confounding factors as possible: (1) received any ocular surgery in the study period, (2) the application of Aspirin (medicine code: A023534100, A024465100, A036599100, A0429341G0, A044016100, A0440161G0, AC37344100, AC373441G0), Clopidogrel (medicine code: AA48649100, AA50126100, AC55026100, AC57819100, BC24863100) and Warfarin (medicine code: B020516100, B023426100) throughout the study period, (3) the presence of ocular trauma (ICD-9 codes: 871.x, 918.x, 921.x) (4) the diagnosis of SCH before the index date which set as the first date of DVDs diagnosis, and (5) age younger than 20 years old or older 100 years. In addition, the individuals in the study group were matched to a non-DVDs individual that serves as the control group, and DVDs patient who could not be matched with a non-DVDs patient were excluded.
2.3. Main Outcome Measurement
The primary outcome in the current study was the development of SCH which with the ICD-9 code of 372.72 and diagnosed by an ophthalmologist (department code: 10) after the index date. Since the acute hemorrhagic conjunctivitis, an inflammatory disease with ocular surface hemorrhage, has a different ICD-9 code in practice (ICD-9 code: 077.4 and 372.30), the possibility to misdiagnose acute hemorrhagic conjunctivitis as SCH by ophthalmologist is minimal. Moreover, we also considered the effect of demographic conditions (i.e., age, gender, urbanization and income level) and the following Charlson comorbidity index (CCI) to standardize the health status in the study population: hypertension (ICD-9 codes: 401–405), diabetes mellitus (DM) (ICD-9 codes: 250.x), acute ischemic heart disease (ICD-9 codes: 410–413), congestive heart failure (ICD-9 codes: 398.91, 402.01, 402.11, 402.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.93, 425.4–425.9, 428.x), peripheral vascular disease (ICD-9 codes: 093.0, 437.3, 440.x, 441.x, 443.1–443.9, 47.1, 557.1, 557.9, V43.4), cerebrovascular disease (ICD-9 codes: 362.34, 430.x–438.x), dementia (ICD-9 codes: 290.x, 294.1, 331.2), chronic pulmonary disease (ICD-9 codes: 416.8, 416.9, 490.x–505.x, 506.4, 508.1, 508.8), rheumatic disease (ICD-9 codes: 446.5, 710.0–710.4, 714.0–714.2, 714.8, 725.x), peptic ulcer disease (ICD-9 codes: 531.x–534.x), liver disease (ICD-9 codes: 070.22, 070.23, 070.32, 070.33, 070.44, 070.54, 070.6, 070.9, 456.0–456.2, 570.x, 571.x, 572.2–572.8, 573.3, 573.4, 573.8, 573.9, V42.7), hemiplegia or paraplegia (ICD-9 codes: 334.1, 342.x, 343.x, 344.0–344.6, 344.9), renal disease (ICD-9 codes: 403.01, 403.11, 403.91, 404.02, 404.03, 404.12, 404.13, 404.92, 404.93, 582.x, 583.0–583.7, 585.x, 586.x, 588.0, V42.0, V45.1, V56.x), malignancy including lymphoma and leukemia, but excluding malignant neoplasm of the skin (ICD-9 codes: 140.x–172.x, 174.x–195.8, 200.x–208.x, 238.6) and coagulation defects (ICD-9-CM codes 286.x). About the ocular co-morbidities, corneal diseases (ICD-9 codes 370.0x, 370.2x, 370.3x, 370.4x, 370.5x, 370.6x, 371.0x, 371.23, 371.4x, 371.6x), cataract (ICD-9 codes 366.10–366.19, 366.8, 366.9), glaucoma (ICD-9 codes 365.x), age-related macular degeneration (AMD) (ICD-9 codes 362.50, 362.51, 362.52), blepharitis (ICD-9 codes 373.0), chronic conjunctivitis including vernal and allergic type (ICD-9 codes 372.1), and noninfectious dermatitis of eyelid including eczematous and allergic type (ICD-9 codes 373.31, 373.32) were considered in the analysis model. We longitudinally traced the data from the index date until the date of SCH diagnosis, withdrawal from the national health insurance program, or 31 December 2013.
2.4. Statistical Analysis
SAS version 9.4 was employed for the analysis. We used the propensity score matching to deal with the potential confounders. The propensity score (predicted probability of DVDs exposure) was estimated for each individual by logistic regression, and the factors included birth year (a difference of age within one year was regarded as the same age), gender, urbanization, income, diseases in the CCI presented as different grade, corneal diseases, cataract, glaucoma, AMD, blepharitis, chronic conjunctivitis and noninfectious dermatitis of eyelid. To standardize the health status more precisely, the percentage of hypertension, DM and renal disease between the two groups were considered separately. The paired DVDs and control individuals were 1:1 matched when the difference of propensity score was nearest between exposure and control. An absolutely standard difference (ASD) was employed to show the differences in the demographic data (age, gender, and income level), and co-morbidities status between the study and control groups after the propensity score matching, the ASD less than 0.1 means the item was balanced between two groups. Then the incidence rate ratio and corresponding 95% confidence intervals (CI) were calculated by Poisson regression. Cox proportional hazard regression that was conducted by including all the demographic data, prominent ocular diseases and systemic co-morbidities mentioned above in the multivariable model that was adopted to compute adjusted hazard ratios (aHR). We plotted Kaplan–Meier curves to indicate the cumulative incidence proportion of SCH between the study and control groups with an interval of four years after the DVDs diagnosis and used the log rank test to determine the significant difference between the survival curves. Since most patients in the NHIRD are Han Taiwanese, race was not considered as a covariate. Results with p < 0.05 were regarded as statistically significant and a p value of less than 0.01 were depicted as p < 0.01.