Linking Health and Social Data to Assess the Performance of High Dimensional Propensity Scores
Bibliographic record
Abstract
IntroductionHigh dimensional propensity scores (HDPS) aim to account for unmeasured confounding. However, it is unclear to what extent HDPS are able to attain this. Objectives and ApproachThis study aimed to test how well HDPS can account for confounding due to social determinants of health when using only health data. A retrospective cohort study was used to examine the effect of exposure to prescription opioids in utero on childhood outcomes (ADHD, school readiness, NICU admission, and hospitalization within the first year of life). Administrative health and social data were linked at the individual level and HDPS for each outcome were calculated using the mothers’ health data. Exposed and unexposed mother-child dyads were then matched. Standardized differences of mothers’ social factors (history of teen birth, lowest income quintile, ever received income assistance (i.e., welfare), ever lived in social housing, history with child protection services, residential mobility, and contact with the justice system) were compared before and after matching to determine to what degree the HDPS could account for differences in social determinants of health. Additional HDPS analyses were performed with social factors included in the HDPS with the health data. ResultsBefore matching, standardized differences between exposed and unexposed groups for the social factors ranged between 0.40-0.75. Income assistance and lowest income quintile consistently had the greatest and smallest standardized difference for all outcomes, respectively. After matching, using health data only, standardized differences decreased considerably, ranging from 0.05-0.27. When including social factors into the HDPS, the addition of income assistance produced the smallest standardized differences with a range of 0.01-0.13 for all outcomes. ConclusionsUsing the HDPS with health data only can reduce confounding due to social factors. If data are available, including income assistance in the HDPS may further reduce confounding for all social determinants of health.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.093 | 0.181 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.004 |
| Bibliometrics | 0.007 | 0.008 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".