Enhancing the power of record linkage involving low quality personal identifiers: use of the best link principle and cause of death prior likelihoods

Quality of routine hospital birth records and the feasibility of their use for creating birth cohorts

BACKGROUND

Linkage between routinely collected hospital birth and other records offers the potential for epidemiological and public health research by developing population-level birth cohorts with cradle-to-grave follow-up. Data from births in English National Health Service hospitals are collected in the Hospital Episode Statistics (HES) database but are of uncertain quality.

METHODS

We examined the range and completeness of birth information recorded in HES and tested an approach for minimizing the effect of hospital-level variations by selecting hospitals with high completeness of recording (≥ 90%) for key fields. We discuss important methodological considerations when using routine healthcare data to develop a birth cohort.

RESULTS

The proportion of missing data in key birth record fields has been decreasing annually, such as gestational age and birth weight (from 46.2 and 43.9% in 2005/06 to 18.1 and 16.9% in 2009/10, respectively). We compared the important characteristics such as size and access to specialist neonatal care between 71 high-coding and 85 low-coding hospitals and found no significant differences, suggesting hospitals with high birth record completeness may be generalizable and representative of all hospitals.

CONCLUSIONS

The completeness of recording of hospital birth information varies greatly between hospitals in England but is improving. It may be preferable and valid to construct cohorts from only hospitals with high completeness of recording.

An empiric modification to the probabilistic record linkage algorithm using frequency-based weight scaling

OBJECTIVE

To incorporate value-based weight scaling into the Fellegi-Sunter (F-S) maximum likelihood linkage algorithm and evaluate the performance of the modified algorithm. Background Because healthcare data are fragmented across many healthcare systems, record linkage is a key component of fully functional health information exchanges. Probabilistic linkage methods produce more accurate, dynamic, and robust matching results than rule-based approaches, particularly when matching patient records that lack unique identifiers. Theoretically, the relative frequency of specific data elements can enhance the F-S method, including minimizing the false-positive or false-negative matches. However, to our knowledge, no frequency-based weight scaling modification to the F-S method has been implemented and specifically evaluated using real-world clinical data.

METHODS

The authors implemented a value-based weight scaling modification using an information theoretical model, and formally evaluated the effectiveness of this modification by linking 51,361 records from Indiana statewide newborn screening data to 80,089 HL7 registration messages from the Indiana Network for Patient Care, an operational health information exchange. In addition to applying the weight scaling modification to all fields, we examined the effect of selectively scaling common or uncommon field-specific values.

RESULTS

The sensitivity, specificity, and positive predictive value for applying weight scaling to all field-specific values were 95.4, 98.8, and 99.9%, respectively. Compared with nonweight scaling, the modified F-S algorithm demonstrated a 10% increase in specificity with a 3% decrease in sensitivity.

CONCLUSION

By eliminating false-positive matches, the value-based weight modification can enhance the specificity of the F-S method with minimal decrease in sensitivity.

Probabilistic record linkage and an automated procedure to minimize the undecided-matched pair problem

Probabilistic record linkage allows the assembling of information from different data sources. We present a procedure when a one-to-one relationship between records in different files is expected but not found. Data were births and infant deaths, 1998-birth cohort, city of São Paulo, Brazil. Pairs for which a one-to-one relationship was obtained and a best-link was found with the highest weight were taken as unequivocally matched pairs and provided information to decide on the remaining pairs. For these, an expected relationship between differences in dates of death and birth registration was found; and places of birth and death registration for neonatal deaths were likely to be the same. Such evidence was used to solve for the remaining pairs. We reduced the number of non-uniquely matched records and of uncertain matches, and increased the number of uniquely matched pairs from 2,249 to 2,827. Future research using record linkage should use strategies from first record linkage runs before a full clerical review (the standard procedure under uncertainty) to efficiently retrieve matches.

A literature review of record linkage procedures focusing on infant health outcomes

Record linkage is a powerful tool in assembling information from different data sources and has been used by a number of public health researchers. In this review, we provide an overview of the record linkage methodologies, focusing particularly on probabilistic record linkage. We then stress the purposes and research applications of linking records by focusing on studies of infant health outcomes based on large data sets, and provide a critical review of the studies in Brazil.

Methodology used for "software for automated linkage in Italy" (SALI)

Linkage of epidemiological registries can provide cost-effective information on the associations between different diseases or exposures in the population under study and on completeness of surveillance system databases. We describe the program SALI (software for automated linkage in Italy) aimed at matching individual records from medium-sized registries (in the order of 100,000 records), where the desired outcome is to miss as few links as possible and, because of low link-likelihood (< 1%), a manual revision of matched pairs is feasible. SALI, developed in CA-Clipper language, uses registry files in dBase format. It requires only name, surname, and date of birth as key fields, and it allows for spelling errors in Italian or other Latin languages through a specific algorithm. Furthermore, a double-blind procedure ensures data confidentiality. The main linkage procedure is based on four stages, two automatic ones, and two where the operator can decide through specific windows whether to accept stage-selected matches. SALI takes into account possible errors in key fields thus reducing false negatives. It was used to solve the problem of linkage between AIDS and cancer registries in Italy. It can be used with every IBM-compatible computer system, assuring uniquely high portability.