Optimal Look-Back Period to Identify True Incident Cases of Diabetes in Medical Insurance Data in the Chinese Population: Retrospective Analysis Study
Bibliographic record
Abstract
BACKGROUND: Accurate estimation of incidence and prevalence is vital for preventing and controlling diabetes. Administrative data (including insurance data) could be a good source to estimate the incidence of diabetes. However, how to determine the look-back period (LP) to remove cases with preceding records remains a problem for administrative data. A short LP will cause overestimation of incidence, whereas a long LP will limit the usefulness of a database. Therefore, it is necessary to determine the optimal LP length for identifying incident cases in administrative data. OBJECTIVE: This study aims to offer different methods to identify the optimal LP for diabetes by using medical insurance data from the Chinese population with reference to other diseases in the administrative data. METHODS: Data from the insurance database of the city of Weifang, China from between January 2016 and December 2020 were used. To identify the incident cases in 2020, we removed prevalent patients with preceding records of diabetes between 2016 and 2019 (ie, a 4-year LP). Using this 4-year LP as a reference, consistency examination indexes (CEIs), including positive predictive values, the κ coefficient, and overestimation rate, were calculated to determine the level of agreement between different LPs and an LP of 4 years (the longest LP). Moreover, we constructed a retrograde survival function, in which survival (ie, incident cases) means not having a preceding record at the given time and the survival time is the difference between the date of the last record in 2020 and the most recent previous record in the LP. Based on the survival outcome and survival time, we established the survival function and survival hazard function. When the survival probability, S(t), remains stable, and survival hazard converges to zero, we obtain the optimal LP. Combined with the results of these two methods, we determined the optimal LP for Chinese diabetes patients. RESULTS: =0.000459), S(t) remained at 0.10, and at 480 days, the frequency of S(t) did not increase. Combining the two methods, we found that the optimal LP is 2 years for Chinese diabetes patients. CONCLUSIONS: The retrograde survival method and CEIs both showed effectiveness. A 2-year LP should be considered when identifying incident cases of diabetes using insurance data in the Chinese population.
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 | 0.009 | 0.014 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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 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".