{"slug":"cross-national","title":"cross-national","summary":"End-to-end cross-national comparison study using KNHANES + NHANES + CHNS (or other parallel surveys). Variable harmonization, parallel weighted analysis, and comparison tables. Supports 2-country (KR+US) and 3-country (KR+US+CN) designs.","platform":"Claude","tags":[],"authorName":"LLM Mart","authorSlug":"llm-mart","score":0,"source":"github","price":null,"verified":false,"createdAt":"2026-09-14T20:48:13.292552Z","repo":{"url":"https://github.com/Aperivue/medsci-skills","stars":318,"forks":75,"license":"MIT","updatedAt":"2026-09-27T05:05:17Z"},"bodyHtml":"<hr>\n<h2>name: cross-national\ndescription: End-to-end cross-national comparison study using KNHANES + NHANES + CHNS (or other parallel surveys). Variable harmonization, parallel weighted analysis, and comparison tables. Supports 2-country (KR+US) and 3-country (KR+US+CN) designs.\ntriggers: cross-national, 한미 비교, Korea US comparison, KNHANES NHANES, 양국 비교, binational, cross-country, 비교연구, 3국 비교, CHNS, 한미중\ntools: Read, Write, Edit, Bash, Grep, Glob\nmodel: opus</h2>\n<h1>Cross-National Comparison Study Skill</h1>\n<p>You are assisting a medical researcher in conducting a cross-national comparison study\nusing parallel nationally representative surveys (e.g., KNHANES for Korea, NHANES for the US, CHNS for China).</p>\n<h2>When to Use</h2>\n<ul>\n<li>Researcher has a clinical question to compare across two countries</li>\n<li>KNHANES + NHANES data available (or other parallel survey pairs)</li>\n<li>Goal: produce a complete analysis with country-stratified results + comparison table</li>\n</ul>\n<h2>Inputs</h2>\n<ol>\n<li><strong>Research question</strong>: exposure → outcome association to compare across countries</li>\n<li><strong>Korean data path</strong>: KNHANES CSV file</li>\n<li><strong>US data path</strong>: NHANES CSV directory (multiple tables to merge)</li>\n<li><strong>Harmonization table</strong> (optional): CSV mapping variables across surveys\n<ul>\n<li>Default: replicate-study skill's <code>harmonization_knhanes_nhanes.csv</code></li>\n</ul>\n</li>\n</ol>\n<h2>Reference Files</h2>\n<ul>\n<li>Harmonization table: <code>medsci-skills/skills/replicate-study/references/harmonization_knhanes_nhanes.csv</code></li>\n<li>Upstream:\n<ul>\n<li><code>medsci-skills/skills/write-paper/references/paper_types/cross_national.md</code> — writing template</li>\n<li><code>medsci-skills/skills/analyze-stats/references/analysis_guides/survey_weighted.md</code></li>\n</ul>\n</li>\n</ul>\n<h2>Workflow</h2>\n<h3>Phase 1: Study Definition</h3>\n<ol>\n<li>Confirm research question: Exposure → Outcome</li>\n<li>Define variable coding for both countries:\n<ul>\n<li>Exposure: PHQ-9, BMI category, smoking, etc.</li>\n<li>Outcome: diabetes, hypertension, mortality, etc.</li>\n<li>Covariates: age, sex, education, income, smoking, alcohol, obesity, CVD</li>\n</ul>\n</li>\n<li>Check harmonization table for variable availability</li>\n<li>Output: study protocol summary for user approval</li>\n</ol>\n<h3>Phase 2: Data Preparation</h3>\n<p><strong>KNHANES (single CSV)</strong>:</p>\n<ol>\n<li>Load CSV, filter age ≥20 (or per protocol)</li>\n<li>Derive variables using KNHANES coding:\n<ul>\n<li>Smoking: BS3_1 (1,2=current, 3=former, 8=never)</li>\n<li>Alcohol: BD1_11 (2-6=frequent, 1=occasional, 8=never)</li>\n<li>Obesity: HE_obe (≥4=obesity for BMI≥25 Asian cutoff)</li>\n<li>PHQ-9: BP_PHQ_1~9, sum score, ≥10=depression</li>\n<li>Diabetes: HE_glu≥126 | HE_HbA1c≥6.5 | DE1_dg=1</li>\n<li>CVD: DI4_dg=1 | DI5_dg=1 | DI6_dg=1</li>\n</ul>\n</li>\n<li>Set survey design: svydesign(id=~psu, strata=~kstrata, weights=~wt_itvex, nest=TRUE)</li>\n</ol>\n<p><strong>NHANES (multiple CSVs)</strong>:</p>\n<ol>\n<li>Load and merge tables by SEQN (DEMO_J, DPQ_J, GHB_J, BIOPRO_J, BMX_J, SMQ_J, ALQ_J, DIQ_J, MCQ_J, BPQ_J)</li>\n<li>Derive variables using NHANES coding:\n<ul>\n<li>Smoking: SMQ020 + SMQ040 (100 cigs + now smoke)</li>\n<li>Alcohol: ALQ121 (past 12 mo frequency → categories)</li>\n<li>Obesity: BMXBMI ≥30 (WHO cutoff, NOT Asian)</li>\n<li>PHQ-9: DPQ010~DPQ090, sum score, ≥10=depression</li>\n<li>Diabetes: LBXSGL≥126 | LBXGH≥6.5 | DIQ010==\"Yes\" (CRITICAL: LBXSGL not LBXSGLU)</li>\n<li>CVD: MCQ160B==\"Yes\" (CHF) | MCQ160C==\"Yes\" (CHD) | MCQ160D==\"Yes\" (angina) | MCQ160E==\"Yes\" (MI)</li>\n<li>HTN: BPXOSY3≥140 | BPXODI3≥90 | BPQ020==\"Yes\"</li>\n</ul>\n</li>\n<li>Set survey design: svydesign(id=~SDMVPSU, strata=~SDMVSTRA, weights=~WTMECPRP, nest=TRUE)</li>\n</ol>\n<h3>Phase 3: Parallel Analysis</h3>\n<p>For EACH country independently:</p>\n<ol>\n<li><strong>Table 1</strong>: Baseline characteristics by exposure (weighted counts + percentages)</li>\n<li><strong>Main analysis</strong>: Sequential logistic regression models\n<ul>\n<li>Model 1 (unadjusted)</li>\n<li>Model 2 (age + sex)</li>\n<li>Model 3 (fully adjusted: + education, income, smoking, alcohol, obesity, CVD)</li>\n</ul>\n</li>\n<li><strong>Subgroup analyses</strong>: By sex, age group, education, income, alcohol, smoking, CVD, obesity</li>\n<li><strong>Dose-response</strong> (if applicable): RCS with 3 knots</li>\n</ol>\n<h3>Phase 4: Cross-National Comparison Table</h3>\n<p>Generate a side-by-side comparison:</p>\n<table>\n<thead>\n<tr>\n<th>Analysis</th>\n<th>Korea wOR (95% CI)</th>\n<th>US wOR (95% CI)</th>\n<th>Direction Agreement</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Overall (fully adjusted)</td>\n<td>...</td>\n<td>...</td>\n<td>✓/✗</td>\n</tr>\n<tr>\n<td>Male</td>\n<td>...</td>\n<td>...</td>\n<td></td>\n</tr>\n<tr>\n<td>Female</td>\n<td>...</td>\n<td>...</td>\n<td></td>\n</tr>\n<tr>\n<td>...</td>\n<td>...</td>\n<td>...</td>\n<td></td>\n</tr>\n</tbody>\n</table>\n<h3>Phase 5: Output Files</h3>\n<pre><code>{working_dir}/\n├── cross_national_report.md    — Study summary + comparison tables\n├── variable_mapping.csv        — Variable mapping with match status\n├── analysis_korea.R            — KNHANES analysis (self-contained)\n├── analysis_us.R               — NHANES analysis (self-contained)\n├── results/\n│   ├── table1_korea.csv\n│   ├── table1_us.csv\n│   ├── main_results_comparison.csv\n│   └── subgroup_comparison.csv\n└── manuscript_draft/           — Optional: Methods + Results draft\n    ├── methods_draft.md\n    └── results_draft.md\n</code></pre>\n<h2>Critical Rules</h2>\n<ol>\n<li><strong>NEVER pool data across countries</strong>. Each country analyzed with its own survey design.</li>\n<li><strong>Country-specific BMI cutoffs</strong>: Korea ≥25 (Asian), US ≥30 (WHO).</li>\n<li><strong>Country-specific income</strong>: KNHANES quartile, NHANES PIR → harmonize to binary.</li>\n<li><strong>Weighted analysis mandatory</strong>: Both KNHANES and NHANES are complex surveys.</li>\n<li><strong>Document all harmonization decisions</strong>: What matches, what needed recoding, what differs.</li>\n<li><strong>Same analytic approach</strong>: Identical model specifications for both countries for fair comparison.</li>\n</ol>\n<h2>KNHANES Variable Coding Reference (validated via Joo 2026 replication)</h2>\n<table>\n<thead>\n<tr>\n<th>Variable</th>\n<th>Raw Var</th>\n<th>Coding</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Smoking</td>\n<td>BS3_1</td>\n<td>1,2=Current; 3=Former; 8=Never</td>\n</tr>\n<tr>\n<td>Alcohol</td>\n<td>BD1_11</td>\n<td>2-6=Frequent (current drinker); 1=Occasional (past-year abstainer); 8=Never</td>\n</tr>\n<tr>\n<td>Obesity</td>\n<td>HE_obe</td>\n<td>1-3=Normal; 4-6=Obesity (BMI≥25)</td>\n</tr>\n<tr>\n<td>Depression</td>\n<td>BP_PHQ_1~9</td>\n<td>Sum ≥10 = depression</td>\n</tr>\n<tr>\n<td>Diabetes</td>\n<td>HE_glu, HE_HbA1c, DE1_dg</td>\n<td>FPG≥126 or HbA1c≥6.5 or DE1_dg=1</td>\n</tr>\n<tr>\n<td>CVD</td>\n<td>DI4_dg, DI5_dg, DI6_dg</td>\n<td>Any = 1 → CVD yes</td>\n</tr>\n<tr>\n<td>Education</td>\n<td>edu</td>\n<td>1-3=Non-college; 4=College</td>\n</tr>\n<tr>\n<td>Income</td>\n<td>incm</td>\n<td>1-3=Bottom 80%; 4=Top 20%</td>\n</tr>\n<tr>\n<td>Survey design</td>\n<td>kstrata, psu, wt_itvex</td>\n<td>strata, cluster, weight</td>\n</tr>\n</tbody>\n</table>\n<h2>NHANES Variable Coding Reference (validated via Joo 2026 cross-national)</h2>\n<p><strong>CRITICAL</strong>: NHANES data downloaded via R <code>nhanesA</code> package uses TEXT LABELS, not numeric codes.</p>\n<table>\n<thead>\n<tr>\n<th>Variable</th>\n<th>Raw Var</th>\n<th>Text Labels → Numeric</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>PHQ-9 items</td>\n<td>DPQ010~DPQ090</td>\n<td>\"Not at all\"→0, \"Several days\"→1, \"More than half the days\"→2, \"Nearly every day\"→3</td>\n</tr>\n<tr>\n<td>Sex</td>\n<td>RIAGENDR</td>\n<td>\"Male\" / \"Female\" (NOT 1/2)</td>\n</tr>\n<tr>\n<td>Smoking (100 cigs)</td>\n<td>SMQ020</td>\n<td>\"Yes\" / \"No\"</td>\n</tr>\n<tr>\n<td>Smoking (now)</td>\n<td>SMQ040</td>\n<td>\"Every day\" / \"Some days\" / \"Not at all\"</td>\n</tr>\n<tr>\n<td>Alcohol freq</td>\n<td>ALQ121</td>\n<td>Text labels (see below)</td>\n</tr>\n<tr>\n<td>Alcohol ever</td>\n<td>ALQ111</td>\n<td>\"Yes\" / \"No\"</td>\n</tr>\n<tr>\n<td>Education</td>\n<td>DMDEDUC2</td>\n<td>5 text levels (see SKILL.md Phase 2)</td>\n</tr>\n<tr>\n<td>Diabetes dx</td>\n<td>DIQ010</td>\n<td>\"Yes\" / \"No\" / \"Borderline\"</td>\n</tr>\n<tr>\n<td>CVD (CHF)</td>\n<td>MCQ160B</td>\n<td>\"Yes\" / \"No\" / \"Don't know\"</td>\n</tr>\n<tr>\n<td>CVD (CHD)</td>\n<td>MCQ160C</td>\n<td>\"Yes\" / \"No\" / \"Don't know\"</td>\n</tr>\n<tr>\n<td>CVD (angina)</td>\n<td>MCQ160D</td>\n<td>\"Yes\" / \"No\" / \"Don't know\"</td>\n</tr>\n<tr>\n<td>Fasting glucose</td>\n<td>LBXSGL (BIOPRO_J)</td>\n<td>Numeric (mg/dL) — note: NOT LBXSGLU</td>\n</tr>\n<tr>\n<td>HbA1c</td>\n<td>LBXGH (GHB_J)</td>\n<td>Numeric (%)</td>\n</tr>\n<tr>\n<td>BMI</td>\n<td>BMXBMI (BMX_J)</td>\n<td>Numeric (kg/m²)</td>\n</tr>\n<tr>\n<td>Weight</td>\n<td>WTMEC2YR (single-cycle) or WTMECPRP (pre-pandemic pooled)</td>\n<td>Numeric</td>\n</tr>\n<tr>\n<td>Strata</td>\n<td>SDMVSTRA</td>\n<td>Numeric</td>\n</tr>\n<tr>\n<td>PSU</td>\n<td>SDMVPSU</td>\n<td>Numeric</td>\n</tr>\n</tbody>\n</table>\n<h3>ALQ121 Text Label Mapping (Alcohol Frequency)</h3>\n<ul>\n<li>Frequent (current drinker): Any specific frequency except \"Never in the last year\"</li>\n<li>Occasional (past-year abstainer): \"Never in the last year\"</li>\n<li>Never (lifetime non-drinker): ALQ111 == \"No\" (ALQ121 will be NA)</li>\n</ul>\n<h3>Additional KNHANES Variables (validated via LE8-Asthma replication)</h3>\n<table>\n<thead>\n<tr>\n<th>Variable</th>\n<th>Raw Var</th>\n<th>Coding</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Asthma</td>\n<td>DJ2_dg</td>\n<td>0=No, 1=Yes (physician dx), 9=Don't know → exclude</td>\n</tr>\n<tr>\n<td>Asthma treatment</td>\n<td>DJ2_pt</td>\n<td>0=No, 1=Yes, 8=N/A, 9=Don't know</td>\n</tr>\n<tr>\n<td>Sleep (2017-18)</td>\n<td>BP16_11/12/13/14</td>\n<td><strong>Clock times, NOT hours!</strong> 11=bed hour, 12=bed min, 13=wake hour, 14=wake min. Calculate: duration = wake_time - bed_time (handle midnight crossing). 99=Don't know→NA</td>\n</tr>\n<tr>\n<td>Sleep (2017-18 weekend)</td>\n<td>BP16_21/22/23/24</td>\n<td>Same format as weekday</td>\n</tr>\n<tr>\n<td>Sleep (2019-20)</td>\n<td>BP16_1/2</td>\n<td>Direct sleep hours (weekday/weekend). 99=Don't know→NA</td>\n</tr>\n<tr>\n<td>PA aerobic</td>\n<td>pa_aerobic</td>\n<td>0=Doesn't meet, 1=Meets guidelines. <strong>Note: values are 0/1, NOT 1/2</strong></td>\n</tr>\n<tr>\n<td>HTN treatment</td>\n<td>DI1_pr</td>\n<td>1=Yes, 0=No (currently treating hypertension)</td>\n</tr>\n<tr>\n<td>Dyslipidemia tx</td>\n<td>DI3_pr</td>\n<td>1=Yes, 0=No (if available)</td>\n</tr>\n<tr>\n<td>Non-HDL chol</td>\n<td>HE_chol - HE_HDL_st2</td>\n<td>Derived: total cholesterol minus HDL</td>\n</tr>\n</tbody>\n</table>\n<h3>Additional NHANES Variables (validated via LE8-Asthma replication)</h3>\n<table>\n<thead>\n<tr>\n<th>Variable</th>\n<th>Raw Var</th>\n<th>Coding</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Asthma</td>\n<td>MCQ010</td>\n<td>\"Yes\" / \"No\" (ever told by doctor)</td>\n</tr>\n<tr>\n<td>Sleep hours</td>\n<td>SLD012</td>\n<td>Numeric (hours/night on weekdays)</td>\n</tr>\n<tr>\n<td>BP treatment</td>\n<td>BPQ020</td>\n<td>\"Yes\" / \"No\" (told by doctor, high BP)</td>\n</tr>\n<tr>\n<td>Cholesterol treatment</td>\n<td>BPQ100D</td>\n<td>\"Yes\" / \"No\" (taking cholesterol Rx)</td>\n</tr>\n<tr>\n<td>PA vigorous work</td>\n<td>PAQ605/PAQ610/PAD615</td>\n<td>Yes/No, days/week, min/day</td>\n</tr>\n<tr>\n<td>PA moderate work</td>\n<td>PAQ620/PAQ625/PAD630</td>\n<td>Yes/No, days/week, min/day</td>\n</tr>\n<tr>\n<td>PA walk/bike</td>\n<td>PAQ635/PAQ640/PAD645</td>\n<td>Yes/No, days/week, min/day</td>\n</tr>\n<tr>\n<td>PA vigorous rec</td>\n<td>PAQ665/PAQ670/PAD675</td>\n<td>Yes/No, days/week, min/day</td>\n</tr>\n<tr>\n<td>PA moderate rec</td>\n<td>PAQ650/PAQ655/PAD660</td>\n<td>Yes/No, days/week, min/day</td>\n</tr>\n<tr>\n<td>Dietary fiber</td>\n<td>DR1TFIBE (DR1TOT_J)</td>\n<td>Numeric (grams, day 1 recall)</td>\n</tr>\n<tr>\n<td>Dietary sodium</td>\n<td>DR1TSODI (DR1TOT_J)</td>\n<td>Numeric (mg)</td>\n</tr>\n<tr>\n<td>Dietary sat fat</td>\n<td>DR1TSFAT (DR1TOT_J)</td>\n<td>Numeric (grams)</td>\n</tr>\n<tr>\n<td>Total energy</td>\n<td>DR1TKCAL (DR1TOT_J)</td>\n<td>Numeric (kcal)</td>\n</tr>\n<tr>\n<td>Total sugars</td>\n<td>DR1TSUGR (DR1TOT_J)</td>\n<td>Numeric (grams)</td>\n</tr>\n<tr>\n<td>Non-HDL chol</td>\n<td>LBXTC - LBDHDD</td>\n<td>Derived: TCHOL_J minus HDL_J</td>\n</tr>\n</tbody>\n</table>\n<h2>CHNS Variable Coding Reference (validated via 3-country batch)</h2>\n<p><strong>Data source</strong>: cpc.unc.edu/projects/china (free registration)\n<strong>Biomarker wave</strong>: 2009 only (N=9,549). Other variables available 1989-2015.\n<strong>Survey design</strong>: No formal weights. Use <code>svydesign(id=~COMMID, weights=~1)</code> or cluster-robust SE.</p>\n<h3>Key Files and Merge Strategy</h3>\n<table>\n<thead>\n<tr>\n<th>File</th>\n<th>Key Variables</th>\n<th>Join Key</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>mast_pub_12</td>\n<td>IDind, GENDER (1=M/2=F), WEST_DOB_Y (birth year)</td>\n<td>IDind</td>\n</tr>\n<tr>\n<td>pexam_00</td>\n<td>HEIGHT, WEIGHT, U10 (waist), SYSTOL1-3, DIASTOL1-3, U22 (HBP dx), U24 (HBP meds), U24A (DM dx), U25 (ever smoked), U27 (still smokes), U40 (alcohol), U41 (freq), U48A (self-health), COMMID</td>\n<td>IDind + filter WAVE==2009</td>\n</tr>\n<tr>\n<td>biomarker_09</td>\n<td>GLUCOSE_MG, HbA1c, TC_MG, TG_MG, HDL_C_MG, LDL_C_MG, HS_CRP, HGB, WBC, ALT, CRE_MG</td>\n<td>IDind</td>\n</tr>\n<tr>\n<td>educ_12</td>\n<td>A12 (education 0-6)</td>\n<td>IDind + filter WAVE==2009</td>\n</tr>\n<tr>\n<td>indinc_10</td>\n<td>indwage (yuan, continuous → quartiles)</td>\n<td>IDind + filter wave==2009</td>\n</tr>\n</tbody>\n</table>\n<h3>Variable Coding</h3>\n<table>\n<thead>\n<tr>\n<th>Variable</th>\n<th>Raw Var</th>\n<th>Coding</th>\n<th>Notes</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Sex</td>\n<td>GENDER</td>\n<td>1=Male, 2=Female</td>\n<td>Same as KNHANES/NHANES</td>\n</tr>\n<tr>\n<td>Age</td>\n<td>WEST_DOB_Y</td>\n<td>age = wave_year - WEST_DOB_Y</td>\n<td>Integer truncation</td>\n</tr>\n<tr>\n<td>BMI</td>\n<td>HEIGHT, WEIGHT</td>\n<td>WEIGHT / (HEIGHT/100)^2</td>\n<td><strong>Obesity: BMI ≥ 28 (WGOC, NOT 25 or 30)</strong></td>\n</tr>\n<tr>\n<td>Waist</td>\n<td>U10</td>\n<td>cm, direct measurement</td>\n<td><strong>Central obesity: ≥90M / ≥80F (IDF-Asian)</strong></td>\n</tr>\n<tr>\n<td>SBP</td>\n<td>SYSTOL1-3</td>\n<td>mean(SYSTOL1, SYSTOL2, SYSTOL3)</td>\n<td>3 readings averaged</td>\n</tr>\n<tr>\n<td>DBP</td>\n<td>DIASTOL1-3</td>\n<td>mean(DIASTOL1, DIASTOL2, DIASTOL3)</td>\n<td>3 readings averaged</td>\n</tr>\n<tr>\n<td>HBP diagnosed</td>\n<td>U22</td>\n<td>0=No, 1=Yes, 9=Don't know (→NA)</td>\n<td></td>\n</tr>\n<tr>\n<td>HBP medication</td>\n<td>U24</td>\n<td>0=No, 1=Yes</td>\n<td></td>\n</tr>\n<tr>\n<td>DM diagnosed</td>\n<td>U24A</td>\n<td>0=No, 1=Yes, 9=Don't know (→NA)</td>\n<td></td>\n</tr>\n<tr>\n<td>Smoking</td>\n<td>U25 + U27</td>\n<td>never(U25<mark>0) / former(U25</mark>1 &amp; U27<mark>0) / current(U25</mark>1 &amp; U27==1)</td>\n<td></td>\n</tr>\n<tr>\n<td>Alcohol</td>\n<td>U40 + U41</td>\n<td>never(U40==0) / occasional(U41≥4) / frequent(U41≤3, ≥1x/week)</td>\n<td>U41: 1=daily, 2=3-4x/wk, 3=1-2x/wk, 4=1-2x/mo, 5=&lt;1x/mo</td>\n</tr>\n<tr>\n<td>Education</td>\n<td>A12</td>\n<td>0=none, 1=primary, 2=lower-mid, 3=upper-mid, 4=technical, 5=university, 6=master+. Recode: 0-2→low, 3-4→mid, 5-6→high</td>\n<td></td>\n</tr>\n<tr>\n<td>Income</td>\n<td>indwage</td>\n<td>Continuous yuan → quartiles within wave</td>\n<td></td>\n</tr>\n<tr>\n<td>Glucose</td>\n<td>GLUCOSE_MG</td>\n<td>mg/dL (also GLUCOSE in mmol/L)</td>\n<td>2009 only</td>\n</tr>\n<tr>\n<td>HbA1c</td>\n<td>HbA1c</td>\n<td>% (direct)</td>\n<td>2009 only</td>\n</tr>\n<tr>\n<td>TC</td>\n<td>TC_MG</td>\n<td>mg/dL</td>\n<td>2009 only</td>\n</tr>\n<tr>\n<td>TG</td>\n<td>TG_MG</td>\n<td>mg/dL</td>\n<td>2009 only</td>\n</tr>\n<tr>\n<td>HDL</td>\n<td>HDL_C_MG</td>\n<td>mg/dL</td>\n<td>2009 only</td>\n</tr>\n<tr>\n<td>hsCRP</td>\n<td>HS_CRP</td>\n<td>mg/L</td>\n<td>2009 only</td>\n</tr>\n<tr>\n<td>Hemoglobin</td>\n<td>HGB</td>\n<td><strong>g/L (divide by 10 for g/dL)</strong></td>\n<td>Unit differs from KR/US</td>\n</tr>\n<tr>\n<td>Self-health</td>\n<td>U48A</td>\n<td>Self-reported health status</td>\n<td>2004-2011</td>\n</tr>\n<tr>\n<td>Depression</td>\n<td>—</td>\n<td><strong>NOT AVAILABLE</strong> in standard download. CES-D exists but needs separate dataset.</td>\n<td>Cannot directly compare with PHQ-9</td>\n</tr>\n</tbody>\n</table>\n<h3>CHNS-Specific Warnings</h3>\n<ol>\n<li><strong>No survey weights</strong>: CHNS is NOT a formally weighted survey. Use unweighted analysis with cluster-robust SE by COMMID. Report as limitation.</li>\n<li><strong>Biomarker = 2009 only</strong>: Glucose, HbA1c, lipids, hsCRP available only in 2009 wave. Other waves lack lab data.</li>\n<li><strong>CES-D not in standard download</strong>: Depression comparison requires separate dataset download from cpc.unc.edu.</li>\n<li><strong>BMI cutoff ≠ KR ≠ US</strong>: China=28, Korea=25, US=30. Use country-specific cutoffs AND sensitivity analysis with WHO cutoff=25.</li>\n<li><strong>SES-health gradient may reverse</strong>: Low education and low income are NOT always risk factors in China (null/protective). This is the \"developing country health transition\" — do NOT treat as a bug.</li>\n<li><strong>Hemoglobin unit</strong>: CHNS reports g/L (KR/US report g/dL). Divide by 10 when comparing.</li>\n<li><strong>Education scale</strong>: 7-level (0-6) vs KR 4-level vs US 5-level. Harmonize to 3-level for comparison.</li>\n</ol>\n<h3>Composite Score Replication Warnings (learned from LE8 replication)</h3>\n<ol>\n<li><strong>BMI cutoff mismatch</strong>: LE8 uses WHO &lt;25 which classifies most Koreans as \"ideal\" → Factor subscore loses BMI discriminatory power in Asian populations. Report this limitation.</li>\n<li><strong>KNHANES sleep = clock times</strong>: BP16_11-14 are bedtime/waketime (hour:min), NOT sleep duration. Must compute <code>wake_time - bed_time</code> with midnight crossing.</li>\n<li><strong>pa_aerobic codes</strong>: Values are 0/1 (not 1/2). Binary → MET-hours approximation is coarse.</li>\n<li><strong>Diet quality scoring</strong>: AHEI-2010 requires detailed food group data; nutrient-based proxy gives different distribution. Recommend downloading NHANES DR1TOT_J for dietary recall nutrients.</li>\n<li><strong>LE8 sensitivity to implementation</strong>: Small scoring differences compound across 8 components → overall score can diverge substantially, especially in the \"moderate\" range where most people cluster.</li>\n</ol>\n<h2>Anti-Hallucination</h2>\n<ul>\n<li><strong>Never fabricate variable names, dataset column names, or variable codings.</strong> If a variable mapping is uncertain, output <code>[VERIFY: variable_name]</code> and ask the user to confirm against the data dictionary.</li>\n<li><strong>Never fabricate statistical results</strong> — no invented p-values, effect sizes, confidence intervals, or sample sizes. All numbers must come from executed code output.</li>\n<li><strong>Never generate references from memory.</strong> Use <code>/search-lit</code> for all citations.</li>\n<li>If a function, package, or API does not exist or you are unsure, say so explicitly rather than guessing.</li>\n</ul>\n","files":[{"path":"SKILL.md","sizeBytes":14912,"isText":true},{"path":"skill.yml","sizeBytes":1408,"isText":true}],"reviewScore":null,"reviewSummary":null,"trust":{"provenance":"trusted-source-unreviewed","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow.","bodySource":null},"bodyLocked":false,"purchaseUrl":null,"sourceUrl":null,"report":{"provenance":"trusted-source-unreviewed","screen":{"ran":true,"outcome":"clean","suspicious":0,"notes":0,"hiddenCharacters":false},"virusScan":{"engine":"clamav","status":"clean","scannedAt":"2026-09-14T20:49:01.18931Z","sha256":"636F7B55F3DD8FABDAB86ACC779145C92BDDBF8C9FA82D0339C7A810E3ECD7C4","sizeBytes":7265},"review":null,"source":{"repositoryUrl":"https://github.com/Aperivue/medsci-skills","path":"skills/cross-national","license":"MIT","commit":"5599b724675a1d788e03cd58dabd3db7c68ca86b","subtreeSha":"648634DF202ED15229207125ACD35D7B8CBA8299FE15821E0DD3C4FBE5EAD0DF","lastSyncedAt":"2026-09-27T19:46:33.449845Z"},"reviewedAt":"2026-09-14T20:54:54.639092Z","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow."},"install":[{"target":"skills-cli","command":"npx skills add https://github.com/Aperivue/medsci-skills/tree/main/skills/cross-national"},{"target":"claude-code","command":"claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install aperivue-medsci-skills@llmmart"},{"target":"git","command":"git clone https://github.com/Aperivue/medsci-skills.git"}]}