{"id":4644,"date":"2026-10-24T22:50:41","date_gmt":"2026-10-24T13:50:41","guid":{"rendered":"https:\/\/best-biostatistics.com\/toukei-er\/?p=4644"},"modified":"2026-10-24T22:50:42","modified_gmt":"2026-10-24T13:50:42","slug":"table1-after-multiple-imputation-and-iptw-with-mice-matchthem-gtsummary","status":"publish","type":"post","link":"https:\/\/best-biostatistics.com\/toukei-er\/entry\/table1-after-multiple-imputation-and-iptw-with-mice-matchthem-gtsummary\/","title":{"rendered":"\u591a\u91cd\u4ee3\u5165\u00d7IPTW\u306eTable 1\u306f\u4f55\u3092\u5e73\u5747\u3059\u308b\uff1fN\u306f\uff1fmice\u30fbMatchThem\u30fbgtsummary\u3067\u4f5c\u308b\u80cc\u666f\u6bd4\u8f03\u8868\u3010R\u3011"},"content":{"rendered":"\n
\n

\u7814\u7a76\u8005<\/strong>: \u6b20\u640d\u5024\u306e\u3042\u308b\u81e8\u5e8a\u30c7\u30fc\u30bf\u3067\u591a\u91cd\u4ee3\u5165\uff08mice\uff09\u3068IPTW\uff08\u50be\u5411\u30b9\u30b3\u30a2\u9006\u78ba\u7387\u91cd\u307f\u4ed8\u3051\uff09\u3092\u4e21\u65b9\u884c\u3046\u5fc5\u8981\u304c\u3042\u308b\u306e\u3067\u3059\u304c\u3001\u4ee3\u5165\u30c7\u30fc\u30bf\u3092\u3069\u3046\u7d71\u5408\u3057\u3066\u60a3\u8005\u80cc\u666f\u6bd4\u8f03\u8868\uff08Table 1\uff09\u3092\u4f5c\u6210\u3057\u3001Excel\u306b\u8cbc\u308c\u3070\u3044\u3044\u306e\u304b\u308f\u304b\u308a\u307e\u305b\u3093\u2026\u2026\u3002<\/p>\n<\/blockquote>\n\n\n\n

\n

\u7d71\u8a08ER<\/strong>: mice<\/code> + MatchThem<\/code>\uff08WeightIt<\/code>\uff09+ gtsummary<\/code> + clipr<\/code> \u3092\u7d44\u307f\u5408\u308f\u305b\u308c\u3070\u3001\u624b\u8a08\u7b97\u3084\u500b\u5225\u30eb\u30fc\u30d7\u51e6\u7406\u3092\u4e00\u5207\u305b\u305a\u30011\u672c\u306e\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u30b3\u30fc\u30c9\u3067\u7d71\u5408\u3055\u308c\u305fTable 1\u3092\u4f5c\u6210\u3057\u3001\u305d\u306e\u307e\u307eExcel\u306e\u30bb\u30eb\u3078\u76f4\u63a5\u30b3\u30d4\u30da\u3067\u304d\u307e\u3059\u3088\uff01\u30dd\u30a4\u30f3\u30c8\u306f\u300c\u4f55\u3092\u5e73\u5747\u3059\u308b\u304b\u300d\u3068\u300cN\u3092\u3069\u3046\u66f8\u304f\u304b\u300d\u3002\u3055\u3063\u305d\u304f\u624b\u9806\u3092\u89e3\u8aac\u3057\u307e\u3057\u3087\u3046\uff01<\/p>\n<\/blockquote>\n\n\n\n

\u81e8\u5e8a\u7814\u7a76\u3084\u89b3\u5bdf\u7814\u7a76\uff08\u30ea\u30a2\u30eb\u30ef\u30fc\u30eb\u30c9\u30c7\u30fc\u30bf\u89e3\u6790\uff09\u306b\u304a\u3044\u3066\u3001\u4ea4\u7d61\u56e0\u5b50\u3092\u8abf\u6574\u3059\u308b\u305f\u3081\u306b\u50be\u5411\u30b9\u30b3\u30a2\u9006\u78ba\u7387\u91cd\u307f\u4ed8\u3051\uff08IPTW: Inverse Probability of Treatment Weighting\uff09\u3092\u5b9f\u65bd\u3059\u308b\u30b1\u30fc\u30b9\u306f\u975e\u5e38\u306b\u591a\u3044\u3002<\/p>\n\n\n\n

\u3057\u304b\u3057\u3001\u5b9f\u969b\u306e\u81e8\u5e8a\u30c7\u30fc\u30bf\u306b\u306f\u80cc\u666f\u56e0\u5b50\uff08\u5e74\u9f62\u30fbBMI\u30fb\u691c\u67fb\u5024\u306a\u3069\uff09\u306b\u300c\u6b20\u640d\u5024\uff08Missing Data\uff09\u300d\u304c\u542b\u307e\u308c\u3066\u3044\u308b\u306e\u304c\u5e38\u3067\u3042\u308b\u3002\u6b20\u640d\u5024\u3092\u5358\u306b\u9664\u5916\u3059\u308b\u300c\u5b8c\u5168\u30b1\u30fc\u30b9\u89e3\u6790\uff08Complete Case Analysis\uff09\u300d\u3092\u884c\u3046\u3068\u3001\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u304c\u5927\u304d\u304f\u6e1b\u5c11\u3057\u3066\u9078\u629e\u30d0\u30a4\u30a2\u30b9\u304c\u751f\u3058\u308b\u305f\u3081\u3001\u67fb\u8aad\u8005\uff08Reviewer\uff09\u304b\u3089\u591a\u91cd\u4ee3\u5165\u6cd5\uff08MI: Multiple Imputation\uff09\u306e\u4f75\u7528\u3092\u6c42\u3081\u3089\u308c\u308b\u5834\u9762\u304c\u5897\u3048\u3066\u3044\u308b\u3002<\/p>\n\n\n\n

\u3053\u3053\u3067\u591a\u304f\u306e\u7814\u7a76\u8005\u304c\u76f4\u9762\u3059\u308b\u306e\u304c\u3001\u300c\u591a\u91cd\u4ee3\u5165\u3067\u88dc\u5b8c\u3055\u308c\u305f\u8907\u6570\u306e\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306b\u5bfe\u3057\u3066IPTW\u3092\u9069\u7528\u3057\u3001\u3069\u3046\u3084\u3063\u3066\u7d71\u5408\u3057\u3066\u60a3\u8005\u80cc\u666f\u6bd4\u8f03\u8868\uff08Table 1\uff09\u3092\u4f5c\u6210\u3059\u308c\u3070\u3088\u3044\u306e\u304b\uff1f\u300d\u3068\u3044\u3046\u5b9f\u88c5\u4e0a\u306e\u5206\u539a\u3044\u58c1\u3067\u3042\u308b\u3002<\/p>\n\n\n\n

\u672c\u8a18\u4e8b\u3067\u306f\u3001\u307e\u305a\u7d71\u5408\u306e\u65b9\u6cd5\u8ad6\uff08\u4f55\u3092\u5e73\u5747\u3057\u3001N\u3092\u3069\u3046\u8868\u793a\u3057\u3001SMD\u3092\u3069\u3046\u51fa\u3059\u304b\uff09<\/strong>\u3092\u6574\u7406\u3057\u305f\u3046\u3048\u3067\u3001mice<\/code>\u30fbMatchThem<\/code>\uff08WeightIt<\/code>\uff09\u30fbcobalt<\/code>\u30fbgtsummary<\/code>\u30fbclipr<\/code> \u3068\u3044\u3046\u6d17\u7df4\u3055\u308c\u305fR\u30d1\u30c3\u30b1\u30fc\u30b8\u3092\u7d44\u307f\u5408\u308f\u305b\u3001\u591a\u91cd\u4ee3\u5165 \u2794 IPTW \u2794 \u7d71\u5408 \u2794 Table 1\u4f5c\u6210 \u2794 Excel\u30b3\u30d4\u30da<\/strong> \u307e\u3067\u3092\u4e00\u6c17\u901a\u8cab\u3067\u5b8c\u7d50\u3055\u305b\u308b\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u30b3\u30fc\u30c9\u3092\u89e3\u8aac\u3059\u308b\u3002<\/p>\n\n\n\n

\u8907\u96d1\u306a\u624b\u8a08\u7b97\u3084\u624b\u5165\u529b\u3092\u5b8c\u5168\u306b\u30bc\u30ed\u306b\u3057\u3001\u78ba\u5b9f\u304b\u3064\u8fc5\u901f\u306b\u8ad6\u6587\u63b2\u8f09\u30ec\u30d9\u30eb\u306e\u80cc\u666f\u56e0\u5b50\u8868\u3092\u4f5c\u308a\u4e0a\u3052\u3088\u3046\u3002<\/p>\n\n\n\n

\u89e3\u6790\u306e\u5168\u4f53\u50cf\uff1a\u306a\u305c\u300c\u591a\u91cd\u4ee3\u5165 \u2794 IPTW \u2794 \u96c6\u8a08\u30fb\u7d71\u5408\u300d\u306e\u9806\u756a\u306a\u306e\u304b\uff1f<\/h2>\n\n\n\n

\u300c\u591a\u91cd\u4ee3\u5165\u3068IPTW\u306f\u3001\u3069\u3061\u3089\u3092\u5148\u306b\u884c\u3046\u3079\u304d\u304b\uff1f\u300d\u3068\u3044\u3046\u8cea\u554f\u306f\u975e\u5e38\u306b\u591a\u3044\u3002\u7d50\u8ad6\u304b\u3089\u8a00\u3048\u3070\u3001\u6b63\u89e3\u306f\u300c\u591a\u91cd\u4ee3\u5165\u3092\u5148\u306b\u884c\u3044\u3001\u88dc\u5b8c\u3055\u308c\u305f\u5404\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u5185\u3067IPTW\u91cd\u307f\u3092\u8a08\u7b97\u3057\u3001\u6700\u5f8c\u306b\u96c6\u8a08\u30fb\u89e3\u6790\u7d50\u679c\u3092\u7d71\u5408\u3059\u308b\u300d\u3068\u3044\u3046\u9806\u5e8f\u3067\u3042\u308b\u3002<\/p>\n\n\n\n

    \n
  1. \u591a\u91cd\u4ee3\u5165\uff08MI\uff09<\/strong>: \u6b20\u640d\u5024\u3092\u542b\u3080\u5143\u30c7\u30fc\u30bf\u304b\u3089\u3001$m$ \u500b\uff08\u4f8b: $m=5$ \u3084 $20$\uff09\u306e\u6b20\u640d\u5024\u88dc\u5b8c\u6e08\u307f\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u4f5c\u6210\u3059\u308b\u3002<\/li>\n\n\n\n
  2. IPTW\u91cd\u307f\u8a08\u7b97<\/strong>: $m$ \u500b\u306e\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u305d\u308c\u305e\u308c<\/strong>\u306b\u304a\u3044\u3066\u3001\u5171\u5909\u91cf\u3092\u7528\u3044\u3066\u50be\u5411\u30b9\u30b3\u30a2\uff08Propensity Score\uff09\u3092\u63a8\u5b9a\u3057\u3001IPTW\u91cd\u307f\u3092\u7b97\u51fa\u3059\u308b\u3002<\/li>\n\n\n\n
  3. \u30d0\u30e9\u30f3\u30b9\u8a55\u4fa1\u30fb\u96c6\u8a08\u306e\u7d71\u5408<\/strong>: \u5404\u4ee3\u5165\u30c7\u30fc\u30bf\u3067\u8a08\u7b97\u3057\u305f\u91cd\u307f\u4ed8\u304d\u96c6\u8a08\u5024\uff08\u80cc\u666f\u56e0\u5b50\u306e\u5e73\u5747\u5024\u3084\u5272\u5408\uff09\u3084\u6a19\u6e96\u5316\u5e73\u5747\u5dee\uff08SMD\uff09\u3092\u3001$m$ \u500b\u306e\u5e73\u5747\u3068\u3057\u3066\u7d71\u5408\u3059\u308b\uff08Rubin’s rules \u306e\u70b9\u63a8\u5b9a\u3068\u540c\u3058\u8003\u3048\u65b9\uff09\u3002<\/li>\n<\/ol>\n\n\n\n

    \u6b20\u640d\u5024\u304c\u3042\u308b\u307e\u307e\u3067\u306f\u3001\u305d\u306e\u60a3\u8005\u306e\u50be\u5411\u30b9\u30b3\u30a2\u3092\u8a08\u7b97\u3067\u304d\u306a\u3044\uff08\u5b8c\u5168\u30b1\u30fc\u30b9\u306b\u9650\u5b9a\u3055\u308c\u3066\u3057\u307e\u3046\uff09\u305f\u3081\u3001\u5148\u306b\u4ee3\u5165\u3057\u3066\u304b\u3089\u50be\u5411\u30b9\u30b3\u30a2\u3092\u63a8\u5b9a\u3059\u308b\u3002\u3055\u3089\u306b\u3001\u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306b\u50be\u5411\u30b9\u30b3\u30a2\u3068\u91cd\u307f\u3092\u8a08\u7b97\u3059\u308b\u300cwithin \u30a2\u30d7\u30ed\u30fc\u30c1\u300d\u304c\u3001\u50be\u5411\u30b9\u30b3\u30a2\u3092\u4ee3\u5165\u9593\u3067\u5e73\u5747\u3057\u3066\u304b\u3089\u91cd\u307f\u3092\u4f5c\u308b\u300cacross \u30a2\u30d7\u30ed\u30fc\u30c1\u300d\u3088\u308a\u3082\u504f\u308a\u304c\u5c0f\u3055\u3044\u3053\u3068\u304c\u3001\u30b7\u30df\u30e5\u30ec\u30fc\u30b7\u30e7\u30f3\u7814\u7a76\u3067\u793a\u3055\u308c\u3066\u3044\u308b\uff08Granger et al., 2019; Leyrat et al., 2019\uff09\u3002<\/p>\n\n\n\n

    Table 1 \u306e\u7d71\u5408\u30eb\u30fc\u30eb\uff1a\u4f55\u3092\u5e73\u5747\u3057\u3001N \u306f\u3069\u3046\u66f8\u304f\u304b\uff1f<\/h2>\n\n\n\n

    Table 1 \u306f\u300c\u52b9\u679c\u306e\u63a8\u5b9a\u300d\u3067\u306f\u306a\u304f\u300c\u80cc\u666f\u306e\u8a18\u8ff0\u300d\u306a\u306e\u3067\u3001\u7d71\u5408\u306b\u5fc5\u8981\u306a\u306e\u306f\u70b9\u306e\u5024\u3060\u3051\u3067\u3042\u308b\u3002\u6a19\u6e96\u8aa4\u5dee\u3092\u5408\u6210\u3059\u308b Rubin’s rules \u306e\u5206\u6563\u306e\u516c\u5f0f\u307e\u3067\u306f\u4f7f\u308f\u306a\u3044\u3002\u6574\u7406\u3059\u308b\u3068\u3001\u6b21\u306e4\u3064\u306e\u30eb\u30fc\u30eb\u306b\u306a\u308b\u3002<\/p>\n\n\n\n

    \u9805\u76ee<\/th>\u7d71\u5408\u306e\u3057\u304b\u305f<\/th>R \u3067\u306e\u6271\u3044<\/th><\/tr><\/thead>
    \u5e73\u5747\u5024\u30fb\u5272\u5408\uff08%\uff09<\/td>\u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306b\u91cd\u307f\u4ed8\u304d\u5024\u3092\u8a08\u7b97\u3057\u3001$m$ \u500b\u3092\u5358\u7d14\u5e73\u5747\u3059\u308b<\/td>\u4ee3\u5165\u30c7\u30fc\u30bf\u3092\u7e26\u306b\u7a4d\u307f\u3001\u91cd\u307f\u3092 $1\/m$ \u500d\u3057\u3066\u96c6\u8a08\uff08survey<\/code> + gtsummary<\/code>\uff09<\/td><\/tr>
    N\uff08\u5217\u898b\u51fa\u3057\uff09<\/td>\u91cd\u307f\u4ed8\u3051\u524d\u306f\u5143\u30c7\u30fc\u30bf\u306e\u5b9f\u969b\u306e\u75c7\u4f8b\u6570<\/strong>\u3002IPTW\u5f8c\u306f\u6709\u52b9\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\uff08ESS\uff09<\/strong>\u306e $m$ \u500b\u306e\u5e73\u5747\u3092\u6dfb\u3048\u308b<\/td>table(df_raw$treat)<\/code> \u3068 cobalt::bal.tab()<\/code> \u306e ESS<\/td><\/tr>
    SMD<\/td>\u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306b\u8a08\u7b97\u3057\u3001$m$ \u500b\u3092\u5e73\u5747\u3059\u308b\uff08\u6700\u5927\u5024\u3082\u78ba\u8a8d\uff09<\/td>cobalt::bal.tab()<\/code> \u306e Mean.Diff<\/td><\/tr>
    \u6b20\u6e2c\u306e\u60c5\u5831<\/td>\u4ee3\u5165\u524d\u306e\u30c7\u30fc\u30bf\u3067\u3001\u5909\u6570\u3054\u3068\u306e\u6b20\u6e2c\u6570\uff08%\uff09\u3092\u5225\u306b\u5831\u544a\u3059\u308b<\/td>tbl_summary(df_raw)<\/code> \u307e\u305f\u306f\u811a\u6ce8<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

    \u4f8b\u3048\u308b\u306a\u3089\u30015\u4eba\u306e\u63a1\u70b9\u8005\uff085\u3064\u306e\u4ee3\u5165\u30c7\u30fc\u30bf\uff09\u304c\u305d\u308c\u305e\u308c\u540c\u3058\u7b54\u6848\u306b\u70b9\u6570\u3092\u3064\u3051\u3001\u6700\u7d42\u6210\u7e3e\u306f\u305d\u306e\u5e73\u5747\u70b9\u306b\u3059\u308b\u3001\u3068\u3044\u3046\u30a4\u30e1\u30fc\u30b8\u3067\u3042\u308b\u3002\u91cd\u307f\u4ed8\u304d\u5e73\u5747\u3092\u300c\u3055\u3089\u306b\u91cd\u307f\u4ed8\u3051\u3057\u3066\u300d\u5e73\u5747\u3057\u76f4\u3059\u5fc5\u8981\u306f\u306a\u3044\u3002<\/p>\n\n\n\n

    N \u306e\u6271\u3044\u306b\u306f\u6ce8\u610f\u304c\u5fc5\u8981\u3067\u3042\u308b\u3002\u4ee3\u5165\u30c7\u30fc\u30bf\u3092\u7e26\u306b\u7a4d\u3080\u3068\u884c\u6570\u306f $m$ \u500d\uff08$m=5$ \u306a\u30891,500\u884c\uff09\u306b\u306a\u308b\u304c\u3001\u75c7\u4f8b\u304c\u5897\u3048\u305f\u308f\u3051\u3067\u306f\u306a\u3044\u3002\u307e\u305f\u3001IPTW\u5f8c\u306e\u300c\u91cd\u307f\u4ed8\u304d\u4eba\u6570\u300d\u306f\u7591\u4f3c\u96c6\u56e3\u306e\u4eba\u6570\u3067\u3042\u308a\u3001\u5b89\u5b9a\u5316\u3057\u3066\u3044\u306a\u3044 ATE \u91cd\u307f\u3067\u306f\u5404\u7fa4\u304c\u5168\u4f53 N \u8fd1\u304f\u307e\u3067\u81a8\u3089\u3080\u3002\u305d\u306e\u305f\u3081\u3001\u91cd\u307f\u4ed8\u3051\u524d\u306f\u5b9f\u969b\u306e\u75c7\u4f8b\u6570\u3001IPTW\u5f8c\u306f\u6709\u52b9\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\uff08ESS\uff09<\/strong>\u3092\u793a\u3059\u306e\u304c\u8aa0\u5b9f\u3067\u3042\u308b\uff08ESS \u306f\u91cd\u307f\u306e\u3070\u3089\u3064\u304d\u306b\u3088\u3063\u3066\u60c5\u5831\u91cf\u304c\u3069\u308c\u3060\u3051\u76ee\u6e1b\u308a\u3057\u305f\u304b\u3092\u8868\u3059\uff09\u3002<\/p>\n\n\n\n

    \uff1e\uff1e\u3082\u3046\u7d71\u8a08\u3067\u60a9\u3080\u306e\u306f\u7d42\u308f\u308a\u306b\u3057\u307e\u305b\u3093\u304b\uff1f\u00a0<\/a><\/strong><\/span><\/p>\r\n\"\"<\/a>\r\n

    \u21911\u4e07\u4eba\u4ee5\u4e0a\u306e\u533b\u7642\u5f93\u4e8b\u8005\u304c\u8cfc\u8aad\u4e2d<\/span><\/strong><\/span><\/p><\/div>

    \u4f7f\u7528\u3059\u308b\u795e\u30d1\u30c3\u30b1\u30fc\u30b85\u9078\u3068\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u306e\u5f79\u5272<\/h2>\n\n\n\n

    \u9ad8\u5ea6\u3067\u8907\u96d1\u306a\u51e6\u7406\u3092\u305f\u3063\u305f1\u672c\u306e\u30b3\u30fc\u30c9\u3067\u30b9\u30de\u30fc\u30c8\u306b\u7e4b\u3050\u305f\u3081\u3001\u672c\u8a18\u4e8b\u3067\u306f\u4ee5\u4e0b\u306e5\u3064\u306e\u30d1\u30c3\u30b1\u30fc\u30b8\uff08\uff0b\u305d\u306e\u571f\u53f0\u3068\u306a\u308b2\u3064\uff09\u3092\u9023\u643a\u3055\u305b\u308b\u3002<\/p>\n\n\n\n

      \n
    • mice<\/code><\/strong>: \u6b20\u640d\u5024\u591a\u91cd\u4ee3\u5165\uff08MICE\u6cd5\uff09\u306e\u6a19\u6e96\u30d1\u30c3\u30b1\u30fc\u30b8\u3002mids<\/code> \u3068\u547c\u3070\u308c\u308b\u591a\u91cd\u4ee3\u5165\u30aa\u30d6\u30b8\u30a7\u30af\u30c8\u3092\u51fa\u529b\u3059\u308b\u3002<\/li>\n\n\n\n
    • MatchThem<\/code><\/strong>\uff08\u4e2d\u8eab\u306e\u8a08\u7b97\u306f WeightIt<\/code><\/strong>\uff09: weightthem()<\/code> \u95a2\u6570\u304c\u4ee3\u5165\u30aa\u30d6\u30b8\u30a7\u30af\u30c8\uff08mids<\/code>\uff09\u3092\u76f4\u63a5\u8aad\u307f\u8fbc\u307f\u3001$m$ \u500b\u306e\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u306b\u4e00\u62ec\u3067IPTW\u91cd\u307f\u3092\u5272\u308a\u5f53\u3066\u308b\u3002WeightIt::weightit()<\/code> \u81ea\u4f53\u306f mids<\/code> \u3092\u53d7\u3051\u53d6\u308c\u306a\u3044\u70b9\u306b\u6ce8\u610f\u3002<\/li>\n\n\n\n
    • cobalt<\/code><\/strong>: \u50be\u5411\u30b9\u30b3\u30a2\u89e3\u6790\u306b\u304a\u3051\u308b\u30d0\u30e9\u30f3\u30b9\u8a55\u4fa1\uff08SMD\u7b49\uff09\u3092\u3001\u4ee3\u5165\u30c7\u30fc\u30bf\u9593\u3067\u7d71\u5408\u30fb\u53ef\u8996\u5316\u3059\u308b\u30d1\u30c3\u30b1\u30fc\u30b8\u3002<\/li>\n\n\n\n
    • gtsummary<\/code><\/strong>\uff08\u571f\u53f0\u306f survey<\/code><\/strong>\uff09: \u91cd\u307f\u4ed8\u304d\u30c7\u30fc\u30bf\uff08survey \u30c7\u30b6\u30a4\u30f3\uff09\u304b\u3089\u3001tbl_svysummary()<\/code> \u3067\u8ad6\u6587\u54c1\u8cea\u306e Table 1\uff08\u80cc\u666f\u6bd4\u8f03\u8868\uff09\u3092\u69cb\u7bc9\u3059\u308b\u3002<\/li>\n\n\n\n
    • clipr<\/code><\/strong>: \u4f5c\u6210\u3057\u305fTable 1\u3092\u3001Excel\u306e\u30bb\u30eb\u5206\u5272\u72b6\u614b\u306e\u307e\u307e\u30af\u30ea\u30c3\u30d7\u30dc\u30fc\u30c9\u3078\u6d41\u3057\u8fbc\u3080\u3002<\/li>\n<\/ul>\n\n\n\n

      \u3010\u30b9\u30c6\u30c3\u30d7\u30d0\u30a4\u30b9\u30c6\u30c3\u30d7\u3011R\u3067\u591a\u91cd\u4ee3\u5165\u00d7IPTW\u306eTable 1\u3092\u4f5c\u6210\u3059\u308b\u5b9f\u8df5\u30b3\u30fc\u30c9<\/h2>\n\n\n\n

      \u3053\u3053\u304b\u3089\u306f\u3001\u5b9f\u969b\u306b\u624b\u5143\u306eR\u74b0\u5883\u3067\u52d5\u4f5c\u3059\u308b\u30b5\u30f3\u30d7\u30eb\u30b3\u30fc\u30c9\u3092\u7528\u3044\u3066\u3001Step 1\u304b\u3089Step 5\u307e\u3067\u9806\u756a\u306b\u89e3\u8aac\u3059\u308b\uff08R 4.5\u3001mice 3.19.0\u3001WeightIt 2.1.0\u3001MatchThem 1.2.1\u3001cobalt 5.0.0\u3001survey 4.5\u3001gtsummary 2.6.1 \u3067\u52d5\u4f5c\u78ba\u8a8d\uff09\u3002<\/p>\n\n\n\n

      \u3042\u3089\u304b\u3058\u3081\u5fc5\u8981\u306a\u30d1\u30c3\u30b1\u30fc\u30b8\u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u30fb\u8aad\u307f\u8fbc\u3093\u3067\u304a\u3053\u3046\u3002<\/p>\n\n\n\n

      # \u5fc5\u8981\u306a\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u8aad\u307f\u8fbc\u307f\nlibrary(tidyverse)\nlibrary(mice)\nlibrary(WeightIt)\nlibrary(MatchThem)\nlibrary(cobalt)\nlibrary(survey)\nlibrary(gtsummary)\nlibrary(clipr)\n<\/code><\/pre>\n\n\n\n
      Step 1: \u6b20\u640d\u5024\u30c7\u30fc\u30bf\u306e\u591a\u91cd\u4ee3\u5165\uff08mice<\/code>\uff09<\/h3>\n\n\n\n
      \u307e\u305a\u306f\u6b20\u640d\u5024\u3092\u542b\u3080\u30b5\u30f3\u30d7\u30eb\u30c7\u30fc\u30bf\u3092\u7528\u610f\u3057\u3001mice()<\/code> \u95a2\u6570\u3067\u591a\u91cd\u4ee3\u5165\u3092\u5b9f\u65bd\u3059\u308b\u3002IPTW\u306e\u52b9\u679c\u304c\u898b\u3048\u308b\u3088\u3046\u306b\u3001\u9ad8\u9f62\u30fb\u9ad8BMI\u30fb\u7537\u6027\u307b\u3069\u6cbb\u7642\u3092\u53d7\u3051\u3084\u3059\u3044\uff08\u4ea4\u7d61\u304c\u3042\u308b\uff09\u30c7\u30fc\u30bf\u306b\u3057\u3066\u3044\u308b\u3002<\/p>\n\n\n\n
      # 1. \u52d5\u4f5c\u78ba\u8a8d\u7528\u306e\u30b5\u30f3\u30d7\u30eb\u30c7\u30fc\u30bf\u3092\u4f5c\u6210\uff08\u6b20\u640d\u5024\u3042\u308a\uff09\nset.seed(123)\nn <- 300\ndf_raw <- tibble(\n  age   = rnorm(n, mean = 65, sd = 10),\n  sex   = factor(sample(c(\"Male\", \"Female\"), n, replace = TRUE)),\n  bmi   = rnorm(n, mean = 24, sd = 4),\n  sbp   = rnorm(n, mean = 135, sd = 15),\n  # \u9ad8\u9f62\u30fb\u9ad8BMI\u30fb\u7537\u6027\u307b\u3069\u6cbb\u7642\u3092\u53d7\u3051\u3084\u3059\u3044\uff08\uff1d\u4ea4\u7d61\u304c\u3042\u308b\uff09\u8a2d\u5b9a\u306b\u3059\u308b\n  treat = rbinom(n, 1, plogis(-0.2 + 0.05 * (age - 65) + 0.10 * (bmi - 24) +\n                                0.5 * (sex == \"Male\")))\n)\n\n# \u610f\u56f3\u7684\u306b\u6b20\u640d\u5024\uff08NA\uff09\u3092\u6ce8\u5165\ndf_raw$bmi[sample(1:n, 40)] <- NA\ndf_raw$sbp[sample(1:n, 25)] <- NA\n\n# 2. mice \u306b\u3088\u308b\u591a\u91cd\u4ee3\u5165\u306e\u5b9f\u884c\uff08\u3053\u3053\u3067\u306f m = 5 \u3068\u3059\u308b\uff09\nimp_data <- mice(df_raw, m = 5, printFlag = FALSE, seed = 123)\nm <- imp_data$m\n<\/code><\/pre>\n\n\n\n
      \u3053\u308c\u3067\u30015\u3064\u306e\u88dc\u5b8c\u6e08\u307f\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u4fdd\u6301\u3059\u308b imp_data<\/code>\uff08mids<\/code> \u30aa\u30d6\u30b8\u30a7\u30af\u30c8\uff09\u304c\u4f5c\u6210\u3055\u308c\u305f\u3002\u306a\u304a\u3001\u672c\u756a\u306e\u89e3\u6790\u3067\u306f $m$ \u3092\u300c\u6b20\u6e2c\u3092\u542b\u3080\u75c7\u4f8b\u306e\u5272\u5408\uff08%\uff09\u300d\u7a0b\u5ea6\u4ee5\u4e0a\u306b\u3059\u308b\u76ee\u5b89\u3082\u3042\u308b\uff08White et al., 2011\uff09\u3002\u3053\u306e\u4f8b\u3067\u306f\u7d042\u5272\u306a\u306e\u3067\u3001$m = 20$ \u7a0b\u5ea6\u306b\u3057\u3066\u3082\u3088\u3044\u3002<\/p>\n\n\n\n
      Step 2: \u4ee3\u5165\u30c7\u30fc\u30bf\u306b\u5bfe\u3059\u308bIPTW\u91cd\u307f\u306e\u8a08\u7b97\uff08MatchThem::weightthem<\/code>\uff09<\/h3>\n\n\n\n
      \u901a\u5e38\u3067\u3042\u308c\u3070 $m$ \u500b\u306e\u30c7\u30fc\u30bf\u3092\u30eb\u30fc\u30d7\u51e6\u7406\u30671\u3064\u305a\u3064\u50be\u5411\u30b9\u30b3\u30a2\u8a08\u7b97\u3059\u308b\u3068\u3053\u308d\u3060\u304c\u3001MatchThem<\/code> \u30d1\u30c3\u30b1\u30fc\u30b8\u306e weightthem()<\/code> \u95a2\u6570\u306f\u3001mice<\/code> \u306e\u51fa\u529b\uff08mids<\/code> \u30aa\u30d6\u30b8\u30a7\u30af\u30c8\uff09\u3092\u305d\u306e\u307e\u307e\u53d7\u3051\u53d6\u308a\u3001\u5185\u90e8\u3067 WeightIt<\/code> \u3092\u547c\u3073\u51fa\u3057\u3066\u8a08\u7b97\u3057\u3066\u304f\u308c\u308b\u3002<\/strong><\/p>\n\n\n\n
      # \u591a\u91cd\u4ee3\u5165\u30c7\u30fc\u30bf\u306b\u5bfe\u3057\u3066\u4e00\u62ec\u3067IPTW\u91cd\u307f\u3092\u7b97\u51fa\nW_imp <- weightthem(\n  treat ~ age + sex + bmi + sbp,\n  datasets  = imp_data,\n  approach  = \"within\",   # \u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306b\u50be\u5411\u30b9\u30b3\u30a2\u3092\u63a8\u5b9a\u3059\u308b\n  method    = \"glm\",      # \u30ed\u30b8\u30b9\u30c6\u30a3\u30c3\u30af\u56de\u5e30\u306b\u3088\u308b\u50be\u5411\u30b9\u30b3\u30a2\u63a8\u5b9a\n  estimand  = \"ATE\",      # \u5168\u4f53\u6cbb\u7642\u52b9\u679c\uff08ATE\uff09\u3092\u30bf\u30fc\u30b2\u30c3\u30c8\n  stabilize = TRUE        # \u5b89\u5b9a\u5316\u91cd\u307f\uff08\u91cd\u307f\u306e\u5408\u8a08\u304c\u5143\u306e N \u4ed8\u8fd1\u306b\u306a\u308b\uff09\n)\n\nW_imp\n<\/code><\/pre>\n\n\n\n
      \u3053\u306e\u308f\u305a\u304b1\u884c\u3067\u30015\u3064\u306e\u4ee3\u5165\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u305d\u308c\u305e\u308c\u306b\u5bfe\u3057\u3066\u500b\u5225\u306b\u50be\u5411\u30b9\u30b3\u30a2\u304c\u8a08\u7b97\u3055\u308c\u3001\u9069\u5207\u306aIPTW\u91cd\u307f\uff08ATE\uff09\u304c\u4ed8\u4e0e\u3055\u308c\u305f\u7d71\u5408\u30aa\u30d6\u30b8\u30a7\u30af\u30c8 W_imp<\/code>\uff08wimids<\/code> \u30aa\u30d6\u30b8\u30a7\u30af\u30c8\uff09\u304c\u5b8c\u6210\u3059\u308b\u3002stabilize = TRUE<\/code> \u306b\u3057\u3066\u304a\u304f\u3068\u3001\u5404\u7fa4\u306e\u91cd\u307f\u306e\u5408\u8a08\u304c\u5143\u306e\u75c7\u4f8b\u6570\u4ed8\u8fd1\u306b\u306a\u308a\u3001Table 1 \u306e\u91cd\u307f\u4ed8\u304d\u4eba\u6570\u304c\u8aad\u307f\u3084\u3059\u304f\u306a\u308b\u3002<\/p>\n\n\n\n
      Step 3: \u30d0\u30e9\u30f3\u30b9\u8a55\u4fa1\u306e\u78ba\u8a8d\uff08cobalt::bal.tab<\/code>\uff09<\/h3>\n\n\n\n
      IPTW\u9069\u7528\u5f8c\u3001\u6cbb\u7642\u7fa4\u3068\u5bfe\u7167\u7fa4\u306e\u9593\u3067\u5171\u5909\u91cf\u30d0\u30e9\u30f3\u30b9\u304c\u5341\u5206\u306b\u4fdd\u305f\u308c\u3066\u3044\u308b\u304b\uff08\u4e00\u822c\u306b\u6a19\u6e96\u5316\u5e73\u5747\u5dee\uff1aSMD < 0.1<\/strong> \u304c\u76ee\u6a19\uff09\u3092\u8a55\u4fa1\u3059\u308b\u3002<\/p>\n\n\n\n
      cobalt<\/code> \u30d1\u30c3\u30b1\u30fc\u30b8\u306e bal.tab()<\/code> \u95a2\u6570\u3092\u4f7f\u7528\u3059\u308b\u3068\u3001\u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306b\u8a08\u7b97\u3057\u305fSMD\u306e\u5e73\u5747\u30fb\u6700\u5927\u5024\u3092\u4e00\u62ec\u3067\u51fa\u529b\u3067\u304d\u308b\u30022\u5024\u5909\u6570\u3082 SMD \u3067\u8868\u793a\u3059\u308b\u305f\u3081\u306b binary = \"std\"<\/code> \u3092\u6307\u5b9a\u3057\u3066\u3044\u308b\uff08\u65e2\u5b9a\u3067\u306f\u5272\u5408\u306e\u5dee\u304c\u305d\u306e\u307e\u307e\u8868\u793a\u3055\u308c\u308b\uff09\u3002<\/p>\n\n\n\n
      # \u8abf\u6574\u524d\u5f8c\u306e\u30d0\u30e9\u30f3\u30b9\uff08SMD\uff09\u3092\u7d71\u5408\u8a55\u4fa1\nbal <- bal.tab(W_imp, un = TRUE, binary = \"std\", abs = TRUE,\n               thresholds = c(m = 0.1))\nbal\n\n# Table 1 \u306b\u8f09\u305b\u308b\u305f\u3081\u3001SMD\uff08\u4ee3\u5165\u30c7\u30fc\u30bf\u9593\u306e\u5e73\u5747\uff09\u3068\u6709\u52b9\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u3092\u53d6\u308a\u51fa\u3059\nsmd_tab <- bal$Balance.Across.Imputations |>\n  rownames_to_column(\"key\") |>\n  select(key, smd_un = Mean.Diff.Un, smd_adj = Mean.Diff.Adj)\ness <- bal$Observations\n<\/code><\/pre>\n\n\n\n
      ## \u51fa\u529b\uff08\u629c\u7c8b\uff09\nBalance summary across all imputations\n               Type Mean.Diff.Un Max.Diff.Un Mean.Diff.Adj Max.Diff.Adj\nprop.score Distance       0.5044      0.5272        0.0044       0.0100\nage         Contin.       0.2990      0.2990        0.0064       0.0100\nsex_Male     Binary       0.1985      0.1985        0.0020       0.0081\nbmi         Contin.       0.3028      0.3478        0.0141       0.0183\nsbp         Contin.       0.0662      0.0836        0.0184       0.0265\nAverage effective sample sizes across imputations\n               0      1\nUnadjusted 141.  159.  \nAdjusted   130.3 150.09\n<\/code><\/pre>\n\n\n\n
      \u7d71\u5408\u3055\u308c\u305f\u8abf\u6574\u5f8cSMD\uff08Mean.Diff.Adj<\/code>\uff09\u3060\u3051\u3067\u306a\u304f\u3001\u4ee3\u5165\u30c7\u30fc\u30bf\u306e\u4e2d\u3067\u6700\u3082\u60aa\u3044\u5024\uff08Max.Diff.Adj<\/code>\uff09\u3082\u3059\u3079\u3066 0.1 \u672a\u6e80\u306b\u53ce\u307e\u3063\u3066\u3044\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3067\u304d\u308c\u3070\u3001\u4ea4\u7d61\u8abf\u6574\u306f\u6210\u529f\u3067\u3042\u308b\u3002\u3053\u306e\u4f8b\u3067\u306f\u3001\u8abf\u6574\u524d\u306b 0.30 \u3042\u3063\u305f\u5e74\u9f62\u3084BMI\u306eSMD\u304c 0.02 \u672a\u6e80\u307e\u3067\u4e0b\u304c\u3063\u3066\u3044\u308b\u3002\u6700\u4e0b\u6bb5\u306e Average effective sample sizes \u304c\u3001Table 1 \u306e\u898b\u51fa\u3057\u306b\u4f7f\u3046\u6709\u52b9\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\uff08ESS\uff09\u3067\u3042\u308b\u3002<\/p>\n\n\n\n
      Step 4: \u7d71\u5408\u30b5\u30de\u30ea\u30fc\u8868\uff08Table 1\uff09\u306e\u4f5c\u6210\uff08survey<\/code> + gtsummary::tbl_svysummary<\/code>\uff09<\/h3>\n\n\n\n
      \u3044\u3088\u3044\u3088\u3001\u8abf\u6574\u5f8c\u306e\u80cc\u666f\u56e0\u5b50\u6bd4\u8f03\u8868\uff08Table 1\uff09\u3092\u4f5c\u6210\u3059\u308b\u3002\u3053\u3053\u304c\u6700\u5927\u306e\u3064\u307e\u305a\u304d\u30dd\u30a4\u30f3\u30c8\u3067\u3001gtsummary::tbl_summary()<\/code> \u306b weightthem()<\/code> \u306e\u7d50\u679c\uff08wimids<\/code>\uff09\u3092\u76f4\u63a5\u6e21\u3059\u3068\u3001\u300cdata \u5f15\u6570\u306f data.frame \u3067\u306a\u3051\u308c\u3070\u306a\u3089\u306a\u3044\u300d\u3068\u3044\u3046\u30a8\u30e9\u30fc\u3067\u6b62\u307e\u308b\u3002<\/strong>\u305d\u3053\u3067\u3001\u6b21\u306e2\u6bb5\u968e\u3067\u4f5c\u308b\u3002<\/p>\n\n\n\n
      4-1. \u4ee3\u5165\u30c7\u30fc\u30bf\u3092\u7e26\u306b\u7a4d\u307f\u3001\u91cd\u307f\u3092 $1\/m$ \u306b\u3059\u308b<\/strong><\/p>\n\n\n\n
      # m \u500b\u306e\u4ee3\u5165\u30c7\u30fc\u30bf\u3092\u7e26\u306b\u7a4d\u307f\u3001\u91cd\u307f\u3092 1\/m \u306b\u3059\u308b\nlong <- complete(W_imp, action = \"long\") |>\n  rename(ipw = weights) |>   # \u5217\u540d weights \u306e\u307e\u307e\u3060\u3068 survey \u306e\u96c6\u8a08\u304c\u8aa4\u308b\u305f\u3081\u6539\u540d\n  mutate(w_un  = 1 \/ m,      # \u91cd\u307f\u4ed8\u3051\u524d\uff1am \u500b\u306e\u5024\u306e\u5e73\u5747\u3060\u3051\u3092\u53d6\u308b\n         w_adj = ipw \/ m)    # IPTW \u5f8c\u3000\uff1a\u91cd\u307f\u4ed8\u304d\u5024\u306e m \u500b\u306e\u5e73\u5747\u3092\u53d6\u308b\n\ndes_un  <- svydesign(ids = ~1, weights = ~w_un,  data = long)\ndes_adj <- svydesign(ids = ~1, weights = ~w_adj, data = long)\n<\/code><\/pre>\n\n\n\n
      \u7a4d\u307f\u4e0a\u3052\u305f\u30c7\u30fc\u30bf\u3067\u3001\u91cd\u307f\u3092 $1\/m$ \u500d\u3057\u3066\u96c6\u8a08\u3059\u308b\u3068\u3001\u300c\u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306e\u91cd\u307f\u4ed8\u304d\u5024\u3092 $m$ \u500b\u5e73\u5747\u3057\u305f\u3082\u306e\u300d\u3068\u540c\u3058\u306b\u306a\u308b\uff08\u5404\u4ee3\u5165\u30c7\u30fc\u30bf\u306e\u91cd\u307f\u306e\u5408\u8a08\u304c\u307b\u307c\u540c\u3058\u3067\u3042\u308b\u305f\u3081\uff09\u3002\u91cd\u307f\u4ed8\u304d\u4eba\u6570\u3082 $m$ \u500d\u306b\u81a8\u3089\u307e\u305a\u30011\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u5206\u306e\u5927\u304d\u3055\u306b\u623b\u308b\u3002<\/p>\n\n\n\n
      \u3053\u3053\u3067\u5217\u540d weights<\/code> \u3092\u5fc5\u305a\u5225\u540d\uff08\u3053\u3053\u3067\u306f ipw<\/code>\uff09\u306b\u5909\u3048\u3066\u304a\u304f<\/strong>\u3002complete()<\/code> \u304c\u8fd4\u3059\u91cd\u307f\u306e\u5217\u540d\u306f weights<\/code> \u3060\u304c\u3001\u3053\u306e\u540d\u524d\u306e\u5217\u304c\u3042\u308b\u3068\u3001survey<\/code> \u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30ab\u30c6\u30b4\u30ea\u5909\u6570\u306e\u96c6\u8a08\uff08svytable()<\/code>\uff09\u304c\u6307\u5b9a\u3057\u305f\u91cd\u307f\u3067\u306f\u306a\u304f\u3053\u306e\u5217\u3092\u4f7f\u3063\u3066\u3057\u307e\u3044\u3001\u30a8\u30e9\u30fc\u3082\u51fa\u306a\u3044\u307e\u307e\u5272\u5408\u304c\u8aa4\u3063\u305f\u5024\u306b\u306a\u308b\uff08survey 4.5 \u3067\u78ba\u8a8d\uff09\u3002<\/p>\n\n\n\n
      4-2. tbl_svysummary()<\/code> \u3067\u8868\u3092\u4f5c\u308a\u3001SMD \u3068 N \u3092\u6574\u3048\u308b<\/strong><\/p>\n\n\n\n
      # \u91cd\u307f\u4ed8\u304d Table 1 \u306b cobalt \u306e SMD \u3092\u7d50\u5408\u3059\u308b\u95a2\u6570\nmake_tbl <- function(des, smd_col) {\n  tbl_svysummary(\n    des,\n    by      = treat, # \u6bd4\u8f03\u7fa4\uff080: \u5bfe\u7167\u7fa4, 1: \u6cbb\u7642\u7fa4\uff09\n    include = c(age, sex, bmi, sbp),\n    statistic = list(\n      all_continuous()  ~ \"{mean} ({sd})\", # \u9023\u7d9a\u5909\u6570\u306f\u5e73\u5747\u5024 (\u6a19\u6e96\u504f\u5dee)\n      all_categorical() ~ \"{n} ({p}%)\"     # \u30ab\u30c6\u30b4\u30ea\u5909\u6570\u306f\u4eba\u6570 (%)\n    ),\n    digits = list(all_continuous() ~ 1, all_categorical() ~ c(0, 1))\n  ) |>\n    modify_table_body(\\(b) b |>\n      mutate(key = if_else(row_type == \"label\", variable,\n                           paste0(variable, \"_\", label))) |>\n      left_join(select(smd_tab, key, SMD = all_of(smd_col)), by = \"key\") |>\n      select(-key)) |>\n    modify_header(SMD ~ \"**SMD**\") |>\n    modify_fmt_fun(SMD ~ \\(x) style_number(x, digits = 2))\n}\n\n# \u5217\u898b\u51fa\u3057\uff1a\u91cd\u307f\u4ed8\u3051\u524d\u306f\u300c\u5b9f\u969b\u306e\u75c7\u4f8b\u6570 N\u300d\u3001IPTW\u5f8c\u306f\u300c\u6709\u52b9\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba ESS\u300d\nn_obs <- table(df_raw$treat)\nt_un <- make_tbl(des_un, \"smd_un\") |>\n  modify_header(stat_1 = sprintf(\"**Control**  \\nN = %d\", n_obs[[\"0\"]]),\n                stat_2 = sprintf(\"**Treated**  \\nN = %d\", n_obs[[\"1\"]]))\nt_adj <- make_tbl(des_adj, \"smd_adj\") |>\n  modify_header(stat_1 = sprintf(\"**Control**  \\nESS = %.1f\", ess[\"Adjusted\", \"0\"]),\n                stat_2 = sprintf(\"**Treated**  \\nESS = %.1f\", ess[\"Adjusted\", \"1\"]))\n\n# \u91cd\u307f\u4ed8\u3051\u524d\u3068 IPTW \u5f8c\u3092\u6a2a\u306b\u4e26\u3079\u308b\ntable1_obj <- tbl_merge(\n  list(t_un, t_adj),\n  tab_spanner = c(\"**Before weighting**\", \"**After IPTW**\")\n) |>\n  modify_footnote_header(\n    sprintf(\"Mean (SD) or n (%%), averaged over %d imputed datasets; weighted n rounded. ESS, effective sample size; SMD, absolute standardized mean difference averaged over imputations.\", m),\n    columns = all_stat_cols(), replace = TRUE\n  )\n\ntable1_obj |> bold_labels() # \u753b\u9762\u8868\u793a\u7528\u306b\u5909\u6570\u540d\u3092\u592a\u5b57\u306b\u3059\u308b\n<\/code><\/pre>\n\n\n\n
      \u3053\u306e\u51e6\u7406\u306b\u3088\u308a\u3001\u5404\u4ee3\u5165\u30c7\u30fc\u30bf\u3067\u306e\u91cd\u307f\u4ed8\u304d\u96c6\u8a08\u7d50\u679c\u304c\u7d71\u5408\u3055\u308c\u3001\u8ad6\u6587\u306b\u305d\u306e\u307e\u307e\u63b2\u8f09\u3067\u304d\u308b\u30d5\u30a9\u30fc\u30de\u30c3\u30c8\u306e\u80cc\u666f\u6bd4\u8f03\u8868\u304c\u51fa\u529b\u3055\u308c\u308b\u3002\u5b9f\u884c\u7d50\u679c\u306f\u6b21\u306e\u3068\u304a\u308a\u3067\u3042\u308b\u3002<\/p>\n\n\n\n
      <\/th>Before weighting<\/th>After IPTW<\/th><\/tr>
      Characteristic<\/th>Control N = 141<\/th>Treated N = 159<\/th>SMD<\/th>Control ESS = 130.3<\/th>Treated ESS = 150.1<\/th>SMD<\/th><\/tr><\/thead>
      age<\/td>63.9 (9.2)<\/td>66.7 (9.5)<\/td>0.30<\/td>65.4 (9.4)<\/td>65.4 (9.3)<\/td>0.01<\/td><\/tr>
      sex<\/td><\/td><\/td><\/td><\/td><\/td><\/td><\/tr>
      Female<\/td>76 (53.9%)<\/td>70 (44.0%)<\/td><\/td>67 (47.7%)<\/td>76 (47.8%)<\/td><\/td><\/tr>
      Male<\/td>65 (46.1%)<\/td>89 (56.0%)<\/td>0.20<\/td>74 (52.3%)<\/td>83 (52.2%)<\/td>0.00<\/td><\/tr>
      bmi<\/td>23.4 (4.0)<\/td>24.6 (4.0)<\/td>0.30<\/td>24.0 (3.9)<\/td>24.0 (4.1)<\/td>0.01<\/td><\/tr>
      sbp<\/td>136.1 (15.2)<\/td>135.0 (16.1)<\/td>0.07<\/td>135.2 (15.3)<\/td>135.5 (16.5)<\/td>0.02<\/td><\/tr><\/tbody><\/table>
      Mean (SD) or n (%), averaged over 5 imputed datasets. ESS, effective sample size; SMD, absolute standardized mean difference averaged over imputations.<\/figcaption><\/figure>\n\n\n\n
      \u91cd\u307f\u4ed8\u3051\u524d\u306f\u5e74\u9f62\u30fb\u6027\u5225\u30fbBMI\u306b\u5dee\uff08SMD 0.20\u301c0.30\uff09\u304c\u3042\u3063\u305f\u304c\u3001IPTW\u5f8c\u306f\u3059\u3079\u3066 SMD 0.02 \u4ee5\u4e0b\u306b\u306a\u3063\u3066\u3044\u308b\u3002IPTW\u5f8c\u306e\u91cd\u307f\u4ed8\u304d\u4eba\u6570\uff08\u4f8b: \u5bfe\u7167\u7fa4 67 + 74 = 141\uff09\u306f\u5143\u306e\u75c7\u4f8b\u6570\u3068\u307b\u307c\u540c\u3058\u3067\u3001\u5b89\u5b9a\u5316\u91cd\u307f\u306e\u52b9\u679c\u304c\u78ba\u8a8d\u3067\u304d\u308b\u3002<\/p>\n\n\n\n
      \u5ff5\u306e\u305f\u3081\u3001\u300c\u7a4d\u307f\u4e0a\u3052\u3066 $1\/m$ \u500d\u300d\u3068\u300c\u4ee3\u5165\u3054\u3068\u306b\u8a08\u7b97\u3057\u3066\u5e73\u5747\u300d\u304c\u4e00\u81f4\u3059\u308b\u304b\u3092\u78ba\u304b\u3081\u3066\u304a\u3053\u3046\u3002<\/p>\n\n\n\n
      # \u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306b\u91cd\u307f\u4ed8\u304d\u5024\u3092\u8a08\u7b97\u3057\u3066\u304b\u3089 m \u500b\u3092\u5e73\u5747\nlong |>\n  group_by(.imp, treat) |>\n  summarise(age  = weighted.mean(age, ipw),\n            male = weighted.mean(sex == \"Male\", ipw), .groups = \"drop\") |>\n  group_by(treat) |>\n  summarise(across(c(age, male), mean))\n\n# \u7a4d\u307f\u4e0a\u3052\u30c7\u30fc\u30bf\u3067\u307e\u3068\u3081\u3066\u91cd\u307f\u4ed8\u304d\u5024\u3092\u8a08\u7b97\nlong |>\n  group_by(treat) |>\n  summarise(age  = weighted.mean(age, ipw),\n            male = weighted.mean(sex == \"Male\", ipw))\n<\/code><\/pre>\n\n\n\n
      \u3069\u3061\u3089\u3082\u5bfe\u7167\u7fa4\u306e\u5e74\u9f62 65.4 \u6b73\u30fb\u7537\u6027\u5272\u5408 0.523\u3001\u6cbb\u7642\u7fa4\u306e\u5e74\u9f62 65.4 \u6b73\u30fb\u7537\u6027\u5272\u5408 0.522 \u3068\u306a\u308a\u3001\u4e00\u81f4\u3059\u308b\u3002<\/p>\n\n\n\n
      Step 5: clipr::write_clip()<\/code> \u3067Excel\u3078\u4e00\u767a\u30b3\u30d4\u30da\uff01<\/h3>\n\n\n\n
      \u4f5c\u6210\u3057\u305f gtsummary<\/code> \u306e\u30aa\u30d6\u30b8\u30a7\u30af\u30c8\u3092\u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0\uff08tibble\uff09\u3078\u5909\u63db\u3057\u3001clipr::write_clip()<\/code> \u3092\u4f7f\u3063\u3066\u30af\u30ea\u30c3\u30d7\u30dc\u30fc\u30c9\u3078\u8ee2\u9001\u3059\u308b\u3002\u898b\u51fa\u3057\u306e\u592a\u5b57\u8a18\u53f7\uff08**<\/code>\uff09\u3084\u6539\u884c\u3001\u7a7a\u6b04\u306e NA<\/code> \u306f\u3001\u8cbc\u308a\u4ed8\u3051\u308b\u524d\u306b\u53d6\u308a\u9664\u3044\u3066\u304a\u304f\u3002<\/p>\n\n\n\n
      # Table 1 \u3092\u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0\u5316\u3057\u3066\u76f4\u63a5\u30af\u30ea\u30c3\u30d7\u30dc\u30fc\u30c9\u306b\u30b3\u30d4\u30fc\ntab_xl <- as_tibble(table1_obj, col_labels = TRUE)\nnames(tab_xl) <- names(tab_xl) |>\n  str_remove_all(\"\\\\*\\\\*\") |>        # \u898b\u51fa\u3057\u306e\u592a\u5b57\u8a18\u53f7 ** \u3092\u5916\u3059\n  str_replace_all(\"\\\\s*\\\\n\", \" \")    # \u898b\u51fa\u3057\u306e\u6539\u884c\u3092\u30b9\u30da\u30fc\u30b9\u306b\ntab_xl[is.na(tab_xl)] <- \"\"          # NA \u3092\u7a7a\u6b04\u306b\u3059\u308b\n\nwrite_clip(tab_xl)\n<\/code><\/pre>\n\n\n\n
      \u3053\u306e\u30b3\u30fc\u30c9\u3092\u5b9f\u884c\u3057\u305f\u3089\u3001\u958b\u3044\u3066\u3044\u308bExcel\u30b7\u30fc\u30c8\u306b\u5207\u308a\u66ff\u3048\u3001\u8cbc\u308a\u4ed8\u3051\u305f\u3044\u30bb\u30eb\u3067 Ctrl + V<\/code>\uff08Mac\u306a\u3089 Cmd + V<\/code>\uff09<\/strong> \u3092\u62bc\u3059\u3060\u3051\u3067\u3088\u3044\u3002<\/p>\n\n\n\n
      \u5404\u6570\u5024\u30fb\u9805\u76ee\u304c\u5b8c\u5168\u306b\u30bb\u30eb\u5206\u5272\u3055\u308c\u305f\u72b6\u614b\u3067Excel\u3078\u7dba\u9e97\u306b\u8cbc\u308a\u4ed8\u3051\u3089\u308c\u308b\u3002\u306a\u304a\u3001write_clip()<\/code> \u306f RStudio \u306a\u3069\u306e\u5bfe\u8a71\u7684\u306a\u30bb\u30c3\u30b7\u30e7\u30f3\u3067\u4f7f\u3046\u95a2\u6570\u3067\u3001Rscript<\/code> \u3067\u30d0\u30c3\u30c1\u5b9f\u884c\u3059\u308b\u3068\u30a8\u30e9\u30fc\u306b\u306a\u308b\u3002\u307e\u305f\u3001\u300cBefore weighting \/ After IPTW\u300d\u306e\u898b\u51fa\u3057\uff08\u30b9\u30d1\u30ca\u30fc\uff09\u306f\u8cbc\u308a\u4ed8\u3051\u3089\u308c\u306a\u3044\u306e\u3067\u3001Excel\u4e0a\u30671\u884c\u8ffd\u52a0\u3057\u3066\u66f8\u304d\u8db3\u305d\u3046\u3002<\/p>\n\n\n\n
      \u81e8\u5e8a\u7814\u7a76\u30fb\u67fb\u8aad\u8005\uff08Reviewer\uff09\u5bfe\u7b56\u306e\u30dd\u30a4\u30f3\u30c8<\/h2>\n\n\n\n
      \n
      \u91cd\u307f\u4ed8\u3051\u5f8c\u306e\u6570\u5024\u306e\u89e3\u91c8\u3068\u8868\u793a\u30de\u30ca\u30fc<\/strong>: IPTW\u9069\u7528\u5f8c\u306e Table 1 \u306b\u304a\u3051\u308b\u9023\u7d9a\u5909\u6570\u306e\u5e73\u5747\u5024\u3084\u30ab\u30c6\u30b4\u30ea\u5909\u6570\u306e\u5272\u5408\uff08%\uff09\u306f\u3001\u91cd\u307f\u4ed8\u3051\u3055\u308c\u305f\u7591\u4f3c\u96c6\u56e3\uff08Pseudo-population\uff09\u306b\u304a\u3051\u308b\u8abf\u6574\u5f8c\u306e\u63a8\u8a08\u5024<\/strong>\u3092\u3001\u4ee3\u5165\u30c7\u30fc\u30bf\u9593\u3067\u5e73\u5747\u3057\u305f\u3082\u306e\u3067\u3042\u308b\u3002\u4eba\u6570\uff08$N$\u3084$n$\uff09\u306e\u8868\u793a\u306b\u3064\u3044\u3066\u306f\u3001\u91cd\u307f\u4ed8\u3051\u524d\u306e\u300c\u5b9f\u969b\u306e\u75c7\u4f8b\u6570\u300d\uff08\u4ee3\u5165\u30c7\u30fc\u30bf\u3092\u7a4d\u307f\u4e0a\u3052\u305f $m$ \u500d\u306e\u6570\u3067\u306f\u306a\u3044\uff09\u3092\u660e\u8a18\u3057\u3001IPTW\u5f8c\u306f\u6709\u52b9\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\uff08Effective Sample Size\uff09\u3092\u898b\u51fa\u3057\u3084\u811a\u6ce8\u306b\u6dfb\u3048\u308b\u306e\u304c\u8aa0\u5b9f\u306a\u4f5c\u6cd5\u3068\u306a\u308b\u3002\u307e\u305f\u3001\u4ee3\u5165\u524d\u306e\u5404\u5909\u6570\u306e\u6b20\u6e2c\u6570\uff08%\uff09\u3082\u3001\u672c\u6587\u30fb\u811a\u6ce8\u30fb\u88dc\u8db3\u8cc7\u6599\u306e\u3044\u305a\u308c\u304b\u3067\u5831\u544a\u3057\u3066\u304a\u304f\u3068\u67fb\u8aad\u3067\u6307\u6458\u3055\u308c\u306b\u304f\u3044\u3002<\/p>\n<\/blockquote>\n\n\n\n
      \u8ad6\u6587\u306e Methods \u6b04\u306b\u66f8\u304f\u82f1\u6587\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u4f8b<\/h3>\n\n\n\n
      \u6295\u7a3f\u8ad6\u6587\u306e\u7d71\u8a08\u89e3\u6790\u624b\u6cd5\uff08Statistical Analysis\uff09\u30bb\u30af\u30b7\u30e7\u30f3\u306b\u8a18\u8f09\u3059\u308b\u82f1\u6587\u30d5\u30ec\u30fc\u30ba\u306e\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u3092\u4ee5\u4e0b\u306b\u793a\u3059\u3002<\/p>\n\n\n\n
      \n
      “Missing covariate values were imputed using multiple imputation by chained equations (MICE) with 5 imputed datasets. Within each imputed dataset, propensity scores were estimated using logistic regression, and stabilized inverse probability of treatment weights (IPTW) for the average treatment effect were calculated (the within approach). Weighted means, proportions, and standardized mean differences (SMDs) were calculated in each imputed dataset and averaged across the imputed datasets. Covariate balance was considered adequate when the absolute SMD was less than 0.1. Effective sample sizes after weighting were also averaged across the imputed datasets.”<\/em><\/p>\n<\/blockquote>\n\n\n\n
      \uff08\u548c\u8a33\uff09\u5171\u5909\u91cf\u306e\u6b20\u6e2c\u5024\u306f\u3001\u9023\u9396\u65b9\u7a0b\u5f0f\u306b\u3088\u308b\u591a\u91cd\u4ee3\u5165\u6cd5\uff08MICE\uff09\u30675\u3064\u306e\u4ee3\u5165\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3092\u4f5c\u6210\u3057\u3066\u88dc\u5b8c\u3057\u305f\u3002\u5404\u4ee3\u5165\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u5185\u3067\u30ed\u30b8\u30b9\u30c6\u30a3\u30c3\u30af\u56de\u5e30\u306b\u3088\u308a\u50be\u5411\u30b9\u30b3\u30a2\u3092\u63a8\u5b9a\u3057\u3001\u5e73\u5747\u51e6\u7f6e\u52b9\u679c\u3092\u5bfe\u8c61\u3068\u3057\u305f\u5b89\u5b9a\u5316\u9006\u78ba\u7387\u91cd\u307f\uff08IPTW\uff09\u3092\u7b97\u51fa\u3057\u305f\uff08within \u30a2\u30d7\u30ed\u30fc\u30c1\uff09\u3002\u91cd\u307f\u4ed8\u304d\u5e73\u5747\u5024\u30fb\u5272\u5408\u30fb\u6a19\u6e96\u5316\u5e73\u5747\u5dee\uff08SMD\uff09\u306f\u4ee3\u5165\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u3054\u3068\u306b\u8a08\u7b97\u3057\u3001\u4ee3\u5165\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u9593\u3067\u5e73\u5747\u3057\u305f\u3002SMD \u306e\u7d76\u5bfe\u5024\u304c 0.1 \u672a\u6e80\u306e\u5834\u5408\u306b\u30d0\u30e9\u30f3\u30b9\u304c\u826f\u597d\u3068\u307f\u306a\u3057\u305f\u3002\u91cd\u307f\u4ed8\u3051\u5f8c\u306e\u6709\u52b9\u30b5\u30f3\u30d7\u30eb\u30b5\u30a4\u30ba\u3082\u4ee3\u5165\u30c7\u30fc\u30bf\u30bb\u30c3\u30c8\u9593\u3067\u5e73\u5747\u3057\u305f\u3002<\/p>\n\n\n\n
      \u3010\u307e\u3068\u3081\u30111\u672c\u3067\u52d5\u304f\u30b3\u30d4\u30da\u7528\u4e00\u62ec\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u30b3\u30fc\u30c9<\/h2>\n\n\n\n
      \u4eca\u56de\u7d39\u4ecb\u3057\u305f\u4e00\u9023\u306e\u51e6\u7406\uff08\u591a\u91cd\u4ee3\u5165 \u2794 IPTW \u2794 \u7d71\u5408 \u2794 Table 1 \u2794 Excel\u51fa\u529b\uff09\u30921\u672c\u306b\u307e\u3068\u3081\u305f\u4e00\u62ec\u30b3\u30fc\u30c9\u3092\u63d0\u793a\u3059\u308b\u3002\u3054\u81ea\u8eab\u306e\u30c7\u30fc\u30bf\u30d5\u30ec\u30fc\u30e0\u540d\u30fb\u5909\u6570\u540d\u306b\u7f6e\u304d\u63db\u3048\u3066\u5b9f\u884c\u3057\u3066\u307b\u3057\u3044\uff08\u6cbb\u7642\u5909\u6570\u306f 0\/1 \u3067\u3001\"0\"<\/code>\u30fb\"1\"<\/code> \u306e\u90e8\u5206\u306f\u7fa4\u306e\u5024\u306b\u5408\u308f\u305b\u3066\u5909\u66f4\u3059\u308b\uff09\u3002<\/p>\n\n\n\n
      # === \u591a\u91cd\u4ee3\u5165 \u00d7 IPTW \u2794 Table 1 \u2794 Excel\u30b3\u30d4\u30da \u4e00\u62ec\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3 ===\nlibrary(tidyverse)\nlibrary(mice)\nlibrary(WeightIt)\nlibrary(MatchThem)\nlibrary(cobalt)\nlibrary(survey)\nlibrary(gtsummary)\nlibrary(clipr)\n\n# 1. \u591a\u91cd\u4ee3\u5165\uff08\u3054\u81ea\u8eab\u306e\u30c7\u30fc\u30bf df_raw \u3068\u6307\u5b9a\u5909\u6570\u306b\u5909\u66f4\uff09\nimp <- mice(df_raw, m = 5, printFlag = FALSE, seed = 123)\nm   <- imp$m\n\n# 2. \u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306b IPTW \u91cd\u307f\u3092\u4e00\u62ec\u8a08\u7b97\uff08\u6cbb\u7642\u5909\u6570 ~ \u5171\u5909\u91cf\uff09\nW <- weightthem(\n  treat ~ age + sex + bmi + sbp,\n  datasets  = imp,\n  approach  = \"within\",\n  method    = \"glm\",\n  estimand  = \"ATE\",\n  stabilize = TRUE\n)\n\n# 3. \u30d0\u30e9\u30f3\u30b9\u8a55\u4fa1\uff08SMD < 0.1 \u306e\u78ba\u8a8d\uff09\u3068\u3001Table 1 \u7528\u306e SMD\u30fbESS \u306e\u53d6\u308a\u51fa\u3057\nbal <- bal.tab(W, un = TRUE, binary = \"std\", abs = TRUE, thresholds = c(m = 0.1))\nbal\nsmd_tab <- bal$Balance.Across.Imputations |>\n  rownames_to_column(\"key\") |>\n  select(key, smd_un = Mean.Diff.Un, smd_adj = Mean.Diff.Adj)\ness <- bal$Observations\n\n# 4. \u4ee3\u5165\u30c7\u30fc\u30bf\u3092\u7e26\u306b\u7a4d\u307f\u3001\u91cd\u307f 1\/m \u306e survey \u30c7\u30b6\u30a4\u30f3\u3092\u4f5c\u308b\nlong <- complete(W, action = \"long\") |>\n  rename(ipw = weights) |>\n  mutate(w_un = 1 \/ m, w_adj = ipw \/ m)\ndes_un  <- svydesign(ids = ~1, weights = ~w_un,  data = long)\ndes_adj <- svydesign(ids = ~1, weights = ~w_adj, data = long)\n\n# 5. \u7d71\u5408 Table 1 \u306e\u4f5c\u6210\nmake_tbl <- function(des, smd_col) {\n  tbl_svysummary(\n    des, by = treat, include = c(age, sex, bmi, sbp),\n    statistic = list(all_continuous() ~ \"{mean} ({sd})\", all_categorical() ~ \"{n} ({p}%)\"),\n    digits    = list(all_continuous() ~ 1, all_categorical() ~ c(0, 1))\n  ) |>\n    modify_table_body(\\(b) b |>\n      mutate(key = if_else(row_type == \"label\", variable, paste0(variable, \"_\", label))) |>\n      left_join(select(smd_tab, key, SMD = all_of(smd_col)), by = \"key\") |>\n      select(-key)) |>\n    modify_header(SMD ~ \"**SMD**\") |>\n    modify_fmt_fun(SMD ~ \\(x) style_number(x, digits = 2))\n}\nn_obs <- table(df_raw$treat)\nt_un  <- make_tbl(des_un, \"smd_un\") |>\n  modify_header(stat_1 = sprintf(\"**Control**  \\nN = %d\", n_obs[[\"0\"]]),\n                stat_2 = sprintf(\"**Treated**  \\nN = %d\", n_obs[[\"1\"]]))\nt_adj <- make_tbl(des_adj, \"smd_adj\") |>\n  modify_header(stat_1 = sprintf(\"**Control**  \\nESS = %.1f\", ess[\"Adjusted\", \"0\"]),\n                stat_2 = sprintf(\"**Treated**  \\nESS = %.1f\", ess[\"Adjusted\", \"1\"]))\ntable1_obj <- tbl_merge(list(t_un, t_adj),\n                        tab_spanner = c(\"**Before weighting**\", \"**After IPTW**\"))\ntable1_obj |> bold_labels()\n\n# 6. \u30af\u30ea\u30c3\u30d7\u30dc\u30fc\u30c9\u3078\u30b3\u30d4\u30fc\ntab_xl <- as_tibble(table1_obj, col_labels = TRUE)\nnames(tab_xl) <- names(tab_xl) |> str_remove_all(\"\\\\*\\\\*\") |> str_replace_all(\"\\\\s*\\\\n\", \" \")\ntab_xl[is.na(tab_xl)] <- \"\"\nwrite_clip(tab_xl)\n\n# \u2794 \u3042\u3068\u306fExcel\u3092\u958b\u3044\u3066 Ctrl + V (Cmd + V) \u3067\u8cbc\u308a\u4ed8\u3051\u308b\u3060\u3051\uff01\n<\/code><\/pre>\n\n\n\n
      \u8907\u96d1\u306b\u898b\u3048\u308b\u9ad8\u5ea6\u306a\u56e0\u679c\u63a8\u8ad6\u3068\u6b20\u640d\u5024\u51e6\u7406\u3082\u3001\u300c\u4ee3\u5165\u30c7\u30fc\u30bf\u3054\u3068\u306b\u8a08\u7b97\u3057\u3066\u5e73\u5747\u3059\u308b\u300d\u300cN \u306f\u5b9f\u969b\u306e\u75c7\u4f8b\u6570\u3068 ESS \u3067\u793a\u3059\u300d\u3068\u3044\u30462\u3064\u306e\u539f\u5247\u3055\u3048\u62bc\u3055\u3048\u308c\u3070\u3001\u6700\u65b0\u306eR\u30d1\u30c3\u30b1\u30fc\u30b8\u7fa4\u3092\u7d44\u307f\u5408\u308f\u305b\u3066\u6570\u5341\u884c\u306e\u30b3\u30fc\u30c9\u3067\u81ea\u52d5\u5316\u3067\u304d\u308b\u3002<\/p>\n\n\n\n
      \u8ee2\u8a18\u30df\u30b9\u3084\u624b\u4f5c\u696d\u306e\u4e0d\u5b89\u3092\u6392\u9664\u3057\u3001\u4fe1\u983c\u6027\u306e\u9ad8\u3044\u7d71\u8a08\u89e3\u6790\u7d50\u679c\u3092\u8ad6\u6587\u3078\u7d44\u307f\u8fbc\u3093\u3067\u3044\u3053\u3046\u3002<\/p>\n\n\n\n
      \u53c2\u8003\u6587\u732e<\/h2>\n\n\n\n