Comparing Multiple Imputation Methods to Address Missing Patient Demographics in Immunization Information Systems: Retrospective Cohort Study
Bibliographic record
Abstract
Background: Immunization Information Systems (IIS) and surveillance data are essential for public health interventions and programming; however, missing data are often a challenge, potentially introducing bias and impacting the accuracy of vaccine coverage assessments, particularly in addressing disparities. Objective: This study aimed to evaluate the performance of 3 multiple imputation methods, Stata's (StataCorp LLC) multiple imputation using chained equations (MICE), scikit-learn's Iterative-Imputer, and Python's miceforest package, in managing missing race and ethnicity data in large-scale surveillance datasets. We compared these methodologies in their ability to preserve demographic distribution, computational efficiency, and performed G-tests on contingency tables to obtain likelihood ratio statistics to assess the association between race and ethnicity and flu vaccination status. Methods: In this retrospective cohort study, we analyzed 2021-2022 flu vaccination and demographic data from the West Virginia Immunization Information System (N=2,302,036), where race (15%) and ethnicity (34%) were missing. MICE, Iterative Imputer, and miceforest were used to impute missing variables, generating 15 datasets each. Computational efficiency, demographic distribution preservation, and spatial clustering patterns were assessed using G-statistics. Results: After imputation, an additional 780,339 observations were obtained compared with complete case analysis. All imputation methods exhibited significant spatial clustering for race imputation (G-statistics: MICE=26,452.7, Iterative-Imputer=128,280.3, Miceforest=26,891.5; P<.001), while ethnicity imputation showed variable clustering patterns (G-statistics: MICE=1142.2, Iterative-Imputer=1.7, Miceforest=2185.0; P: MICE<.001, Iterative-Imputer=1.7, Miceforest<.001). MICE and miceforest best preserved the proportional distribution of demographics. Computational efficiency varied, with MICE requiring 14 hours, Iterative Imputer 2 minutes, and miceforest 10 minutes for 15 imputations. Postimputation estimates indicated a 0.87%-18% reduction in stratified flu vaccination coverage rates. Overall estimated flu vaccination rates decreased from 26% to 19% after imputations. Conclusions: Both MICE and Miceforest offer flexible and reliable approaches for imputing missing demographic data while mitigating bias compared with Iterative-Imputer. Our results also highlight that the imputation method can profoundly affect research findings. Though MICE and Miceforest had better effect sizes and reliability, MICE was much more computationally and time-expensive, limiting its use in large, surveillance datasets. Miceforest can use cloud-based computing, which further enhances efficiency by offloading resource-intensive tasks, enabling parallel execution, and minimizing processing delays. The significant decrease in vaccination coverage estimates validates how incomplete or missing data can eclipse real disparities. Our findings support regular application of imputation methods in immunization surveillance to improve health equity evaluations and shape targeted public health interventions and programming.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.002 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".