sql_mini_mentalhealth
Focal cohort: n=15,780 (Adults 18–25).
Variables: AMHTXND2, IRWRKSTAT.
Full-time employment drops ~67% → ~61% when unmet need is present.
Unmet mental health need = measurable, catastrophic workforce detachment.
Every phase of this study maps directly to the skills healthcare analytics organizations depend on. Defining a testable hypothesis from a complex federal dataset mirrors the clinical question framing required for HEDIS measure analysis and CMS quality reporting. Building ETL pipelines to resolve schema drift across NSDUH survey years is the same discipline applied to normalizing inconsistent EHR exports, payer claim formats, and ICD-10/CPT code structures across systems. Chi-Square independence testing is directly applicable to denial pattern analysis—determining whether denial rate is statistically dependent on payer type, service line, or coding practice. Translating codebook variables into readable nomenclature for non-technical stakeholders is exactly what RCM teams need when surfacing A/R aging trends or clean claim rates to CFO-level audiences. The full arc demonstrated here—question, pipeline, validation, insight, communication—is the complete workflow of a productive healthcare data analyst.
PROJECT OVERVIEW
The Story Behind The Data
This analysis approaches a deceptively simple question with rigorous statistical methodology: does failing to treat mental health needs result in measurable workforce detachment? Using the 2019 SAMHSA NSDUH dataset, the study isolates 15,780 adults aged 18–25 and tests for statistical dependence between unmet mental health need and employment status.
The hypothesis is that adults reporting an unmet need will have statistically higher odds of unemployment or labor force detachment—controlling for age, sex, race/ethnicity, and education level.
PRIMARY_VISUALIZATION // UNMET NEED VS EMPLOYMENT
Unmet mental health need stratified by sex and employment outcome — Chi-Square p < 0.001
METHODOLOGY
The Analytical Framework
- ASK: Defined the target population: adults aged 18–25, unmet mental health need as exposure, employment status as outcome.
- PREPARE: Extracted 2019 NSDUH PUF data; renamed raw codebook variables (AMHTXND2, IRWRKSTAT) into readable schema for downstream SQL and Tableau.
- PROCESS: Built PostgreSQL pipelines to resolve schema drift across survey years and enforce statistical weight integrity.
- ANALYZE: Executed Chi-Square test of independence on focal cohort (n = 15,780), confirming statistical significance at p < 0.001.
DATA_GOVERNANCE // VARIABLE MAPPING
| Raw Syntax | Cleaned Output | Data Dictionary Rationale |
|---|---|---|
| AMHTXND2 | unmet_need | Perceived unmet need for mental health treatment. |
| IRWRKSTAT | employment_status | Revised imputed employment status. |
| IRWRKSTAT18 | employment_status_18_plus | Age-restricted cohort analysis base. |
| AMHSVTYP | mental_health_treatment_type | Type of mental health treatment received. |
| SPDYRADJ | serious_psychological_distress | Standardized variable mapping across years. |
| catage / AGE | age_group | Unified age categorization across TEDS-A and NSDUH pipelines. |
| PSYPROB | cooccurring_mental_substance_disorder | Makes clinical correlation explicit. |
| EDUC | education_level | Education at time of admission/survey. |
| PRIMINC | primary_income_source | Principal source of income/support. |
| DETNLF | not_in_labor_force_detail | Detail matrix for labor force detachments. |
RESULTS ANALYSIS
What The Data Shows Us
- Contingency & Cross-Tabulation: Within populations with unmet mental health needs, the percentage capable of holding full-time employment drops from ~67% down to ~61%, shifting heavily toward part-time or total unemployment.
- Chi-Square Test of Independence: Statistically proves (Rejecting the Null Hypothesis) that employment status is intrinsically dependent on mental health treatment accessibility.
- Modeling Progression: Initial logistic regression failed because both primary vectors were categorical integers with no continuous scale — validating the necessity of Chi-Square over OLS.
TL;DR
Results & Findings
When people do not have an unmet mental health need, the vast majority are securely engaged in full-time work. When people do suffer from an unmet mental health need, the rate of total unemployment and severe workforce detachment spikes substantially relative to population size.
COHORT: n = 15,780 (Adults 18–25) | P-VALUE: < 0.001
INTERPRETATION: Unmet mental health need = measurable, catastrophic workforce detachment.