BMC oral health · 2026
Maxillary sinus-related adverse event reports associated with endosseous dental implants in the FDA MAUDE database: a retrospective text-mining analysis.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Liu Cheng, Tang Ting, Chen Yichen, Meng Xin, Song Zhou, Liu Dongsheng, Zhu Lin
Аннотация
BACKGROUND: Maxillary sinus-related adverse events are clinically relevant in posterior maxillary implant therapy, but their reporting patterns in public postmarket surveillance data remain incompletely characterized.
METHODS: This retrospective text-mining study analyzed FDA Manufacturer and User Facility Device Experience (MAUDE) reports from 2015 to 2025 involving endosseous dental implants with product code DZE. Device records were linked to FOITEXT (MAUDE narrative text) records and Master Event fields by MDR_REPORT_KEY. Sinus-related narratives were identified using staged keyword rules, refined into high-specificity candidates, classified into event categories, assessed for duplicate and template-like text, and validated using independent blinded dual review of a stratified quality-control sample.
RESULTS: Among 3,386,630 unique DZE reports, the workflow retained 6,386 high-specificity maxillary sinus-related candidates. Independent blinded dual review of a 449-record stratified validation sample showed substantial agreement for inclusion status (Cohen's κ = 0.778) and almost perfect agreement for event-category assignment (Cohen's κ = 0.907). The conservative consensus-confirmed positive predictive value (PPV) was 96.88% (435/449). Sinus membrane perforation narratives were the predominant rule-based report category, but this pattern should be interpreted as a reporting and documentation pattern rather than a true clinical frequency. Unique-text sensitivity analysis reduced repeated narratives but did not materially change the category-level pattern.
CONCLUSIONS: This study characterizes maxillary sinus-related reporting patterns in a passive postmarket surveillance database. The findings highlight the value and limitations of transparent text-mining workflows for implant-related adverse event surveillance and should not be interpreted as clinical incidence, comparative risk, manufacturer performance, or causality.
METHODS: This retrospective text-mining study analyzed FDA Manufacturer and User Facility Device Experience (MAUDE) reports from 2015 to 2025 involving endosseous dental implants with product code DZE. Device records were linked to FOITEXT (MAUDE narrative text) records and Master Event fields by MDR_REPORT_KEY. Sinus-related narratives were identified using staged keyword rules, refined into high-specificity candidates, classified into event categories, assessed for duplicate and template-like text, and validated using independent blinded dual review of a stratified quality-control sample.
RESULTS: Among 3,386,630 unique DZE reports, the workflow retained 6,386 high-specificity maxillary sinus-related candidates. Independent blinded dual review of a 449-record stratified validation sample showed substantial agreement for inclusion status (Cohen's κ = 0.778) and almost perfect agreement for event-category assignment (Cohen's κ = 0.907). The conservative consensus-confirmed positive predictive value (PPV) was 96.88% (435/449). Sinus membrane perforation narratives were the predominant rule-based report category, but this pattern should be interpreted as a reporting and documentation pattern rather than a true clinical frequency. Unique-text sensitivity analysis reduced repeated narratives but did not materially change the category-level pattern.
CONCLUSIONS: This study characterizes maxillary sinus-related reporting patterns in a passive postmarket surveillance database. The findings highlight the value and limitations of transparent text-mining workflows for implant-related adverse event surveillance and should not be interpreted as clinical incidence, comparative risk, manufacturer performance, or causality.