68-2

29/110

PatientsWe accessed the hospital medical record system and identified all adult male or non-pregnant female patients who visited any outpatient depart-ment of Juntendo University Hospital - a 1051-bed university-affiliated hospital in Tokyo, Japan, and who were diagnosed with HZ between October, 2018 and February, 2020. Among them, we investi-gated patients who were treated with either AMNV or FCV. Excluded were patients who did not complete the initial 7-day session of the anti-herpetic agent therapy for any reason, and patients whose pain scores evaluated on an 11-point numer-ical rating scale (NRS) (0 = no pain, 10 = worst pain) during the acute HZ period were not thor-oughly recorded on medical records.Data collectionFrom the medical record, we collected data on baseline clinical characteristics, such as patients’ demography, including age and sex; antiherpetic agents administered, including AMNV and FCV; the time from onset of symptoms/signs of HZ to the initiation of antiherpetic agent therapy; patients’ conditions prior to onset of HZ, including adminis-tration of systemic antibiotics for any bacterial infection, daily uses of systemic steroids for under-lying diseases, other immunocompromised states, and daily uses of antidepressants; and location of rashes. Immunocompromised patients other than steroid users were defined as patients receiving chemotherapy and/or radiotherapy for cancer. Although antidepressants have not been listed as a predisposing factor for HZ, we included them in clinical characteristics because they might affect intensities of acute HZ pain through their well-known analgesic effects on PHN1). We also collected data on primary clinical outcomes associated with intensities of acute HZ pain, including non-steroidal anti-inflammatory drugs (NSAIDs) required to relieve acute HZ pain; the NRS pain score reported by patients at Day 0 just prior to the initiation of antiherpetic agent therapy, and NRS pain scores at Day 3-4, Day 7, and Week 2-3 after initiating the therapy. Further, we collected data on secondary clinical outcomes associated with development of PHN as defined as pain lasting for 6 months or more7-9), including the incidence of PHN and predis-posing factors of PHN. Study designTo compare effects of AMNV and FCV on primary clinical outcomes, we first compared NRS pain scores of acute HZ pain and requirements of NSAIDs between patients treated with AMNV and those treated with FCV. We then evaluated the effect of the time from onset of HZ symptoms/signs to the initiation of antiherpetic agent therapy by comparing the primary clinical outcomes between early visi-tors receiving the therapy within 3 days after onset and late visitors receiving it later than 3 days, considering that the early initiation of the antiher-petic agents within 72 hours has been recom-mended to achieve better clinical outcomes5, 6, 10, 11). We also conducted subgroup analyses by comparing the primary clinical outcomes between AMNV and FCV separately in early visitors and in late visitors. At the same time, we compared the secondary clinical outcomes including the incidence of PHN between antiherpetic agents and between early and late visitors. We also attempted to identify predictors of PHN.Statistical analysisVariables are, in principle, shown as the Mean ± SD, Median (Interquartile Range), or Number (%) according to data types. Parametric data such as age were compared between dichotomized patient groups using the unpaired t test. Nonparametric data such as NRS pain scores were compared between the groups with the Mann-Whitney U test. Changes in pain scores were examined with the Wilcoxon signed-rank test followed by the Bonferroni correction. Categorical data were compared between the groups with the Pearson’s chi-square test. Comparisons of clinical outcomes between patient groups were performed also after the nearest neighbor propensity score matching (PSM) in 1:1 ratio was applied to generate a propensity score-matched (PSM) pair of patients with comparable clinical characteristics, using scores calculated with a multivariate logistic regression model based on differences in clinical characteris-tics. The univariate logistic regression analysis was used to identify predictors of PHN. The receiv-er-operating characteristic (ROC) analysis was used to obtain a cutoff point to construct categor-ical data from continuous data. The statistical anal-ysis was performed with StatFlex ver. 7 (ARTECH, 121