Search

Objective To identify data linkage errors in the form of possible false matches, where two patients appear to share the same unique identification number. Data Source Hospital Episode Statistics ( HES) in England, United Kingdom. Study Design Data on births and re-admissions for infants (April 1, 2011 to March 31, 2012; age 0-1 year) and adolescents (April 1, 2004 to March 31, 2011; age 10-19 years). Data Collection/Extraction Methods Hospital records pseudo-anonymized using an algorithm designed to link multiple records belonging to the same person. Six implausible clinical scenarios were considered possible false matches: multiple births sharing HESID, re-admission after death, two birth episodes sharing HESID, simultaneous admission at different hospitals, infant episodes coded as deliveries, and adolescent episodes coded as births. Principal Findings Among 507,778 infants, possible false matches were relatively rare ( n = 433, 0.1 percent). The most common scenario (simultaneous admission at two hospitals, n = 324) was more likely for infants with missing data, those born preterm, and for Asian infants. Among adolescents, this scenario ( n = 320) was more common for males, younger patients, the Mixed ethnic group, and those re-admitted more frequently. Conclusions Researchers can identify clinically implausible scenarios and patients affected, at the data cleaning stage, to mitigate the impact of possible linkage errors. [ABSTRACT FROM AUTHOR]

Background Linkage of electronic healthcare records is becoming increasingly important for research purposes. However, linkage error due to mis-recorded or missing identifiers can lead to biased results. We evaluated the impact of linkage error on estimated infection rates using two different methods for classifying links: highest-weight (HW) classification using probabilistic match weights and prior-informed imputation (PII) using match probabilities. Methods A gold-standard dataset was created through deterministic linkage of unique identifiers in admission data from two hospitals and infection data recorded at the hospital laboratories (original data). Unique identifiers were then removed and data were re-linked by date of birth, sex and Soundex using two classification methods: i) HW classification - accepting the candidate record with the highest weight exceeding a threshold and ii) PII-imputing values from a match probability distribution. To evaluate methods for linking data with different error rates, non-random error and different match rates, we generated simulation data. Each set of simulated files was linked using both classification methods. Infection rates in the linked data were compared with those in the gold-standard data. Results In the original gold-standard data, 1496/20924 admissions linked to an infection. In the linked original data, PII provided least biased results: 1481 and 1457 infections (upper/lower thresholds) compared with 1316 and 1287 (HW upper/lower thresholds). In the simulated data, substantial bias (up to 112%) was introduced when linkage error varied by hospital. Bias was also greater when the match rate was low or the identifier error rate was high and in these cases, PII performed better than HW classification at reducing bias due to falsematches. Conclusions This study highlights the importance of evaluating the potential impact of linkage error on results. PII can help incorporate linkage uncertainty into analysis and reduce bias due to linkage error, without requiring identifiers. [ABSTRACT FROM AUTHOR]

Objectives: Interventions to reduce hospital-acquired bloodstream infection have succeeded in reducing rates in U.S. PICUs, but there is a lack of evidence for the impact of similar interventions in the United Kingdom. We assessed variation in bloodstream infection rates within and between PICUs over a 10-year period, during which time infection control strategies (care bundles) were implemented. Design: Observational study linking laboratory data to national audit data of pediatric intensive care admissions (Paediatric Intensive Care Audit Network). Setting: Twenty PICUs in England and Wales, 2003-2012. Patients: One hundred and two thousand nine hundred ninetynine children less than 16 years. Interventions: Implementation of infection control strategies in PICU captured through a survey of clinicians. Measurements and Main Results: Rates of bloodstream infection per 1,000 bed-days were estimated from samples taken between 2 days after admission and up to 2 days following discharge from PICU. Two percent of children experienced at least one bloodstream infection, corresponding to 5.11 (95% CI, 4.90-5.31) per 1,000 bed-days. There was a significant difference in trends preimplementation of infection control strategies (annual decrease of 8.0%; 95% CI, 6.3-9.7%) versus postimplementation (annual decrease of 13.4%; 95% CI, 10.3-16.4%). By 24 months postimplementation, the rate of bloodstream infection had fallen 25.5% and was 15.1% lower than would have been expected if preimplementation trends had continued. Conclusions: Our population-based study of PICUs in England and Wales demonstrates a steady decline in bloodstream infection rates over time. In addition, there was a significant and incremental further decrease in rates associated with timing of implementation of infection control strategies. Assessment of bloodstream infection trends before as well as after implementation of infection control strategies can be facilitated using data linkage and is important to avoid overestimating the impact of unitlevel interventions to improve infection control. Advances in collection and linkage of real-time data could further support quality improvement efforts. [ABSTRACT FROM AUTHOR]