Duplicate Patient Records – Implication for Missed Laboratory Results

Abstract

Introduction:

Although duplicate records are a potential patient safety hazard, the actual clinical harm associated with these records has never been studied. We hypothesized that duplicate records will be associated with missed abnormal laboratory results.

Methods:

A retrospective, matched, cohort study of 904 events of abnormal laboratory result (HgbA1c, TSH, Vitamin B₁₂, LDL). We compared the rates of missed laboratory results between patients with duplicate and non-duplicate records from the ambulatory clinics. Cases were matched according to test and ordering physician.

Results:

Duplicate records were associated with a higher rate of missed laboratory results (OR=1.44, 95% CI 1.1–1.9). Other factors associated with missed lab results were tests performed as screening (OR=2.22, 95% CI 1.4–3.4), and older age (OR=1.15 for every decade, 95% CI 1.01–1.2). In most cases test results were reported into the main patient record.

Discussion:

Duplicate records were associated with a higher risk of missing important laboratory results.

Introduction

Duplicate patient records occur when a single patient is associated with more than one record. In such a case, pertinent patient information may be fragmented between two or more records which in turn could lead to patient harm and considerable costs. ¹ For example a physician may not be aware of a drug allergy and prescribe a dangerous drug, or an important laboratory result may be missed because it is stored in the other record. Interestingly, while duplicate records are a widely recognized problem, there is little empirical evidence regarding their potential clinical harm. ^2,3 We hypothesized that abnormal test results were more likely to be missed when a single patient had more than one record.

Background

Duplicate records may be generated by technical and administrative processes such as errors in entering patient information, or the integration of patient data from different information systems.¹ The associated cost of a single pair of duplicate records has been estimated at $50. ² This reflects administrative costs, such as time spent to locate the correct record, to re-collect patient data present in a duplicate record, the cost of supplies to support the duplicate records, problems with billing, repeated testing and time spent to reconcile records.² These costs are likely to increase with increasing implementation of “meaningful use”. ¹

Large as these costs may be, they seem negligible when compared to the potential patient harm associated with delayed treatment due to a provider’s inability to access the complete patient record.⁴ For example, clinicians fail to follow up on 7–65% of laboratory tests in ambulatory care.⁵ This has been associated with delays in diagnosis and treatment that may cause significant morbidity and mortality. ^4,6 In the presence of duplicate records, it is reasonable to assume that the rate of missed laboratory results will increase as these may not be stored in the record the physician is reviewing. The purpose of this study was to compare the rate of missed laboratory results between duplicate and non-duplicate records as a marker for the potential harm associated with duplicate records. We compared the rates of missed abnormal laboratory results between duplicate and non-duplicate records, regardless of whether a laboratory result was stored in the main record or in the secondary record. Our rationale was that, the existence of two records in itself is likely to disrupt the workflow and cause errors.

Methods

We conducted a retrospective, matched, cohort study of abnormal laboratory results. Records were retrieved from the clinical data warehouse (CDW) of the ambulatory clinics at the University of Texas Health Science Center in Houston. The CDW currently contains clinical data on over 400,000 patients. The CDW is separated from the operational electronic health record (EHR) system and uses a master patient index (MPI) to list all records associated with individual patients.

We selected only duplicate records that had some type of clinical data in both records (i.e., clinical notes, laboratory results, imaging studies, etc.). Our rationale was that records with clinical data were more likely to have been accessed by healthcare personnel at some point, and therefore have a greater potential of being clinically significant. We distinguished between the primary record (the main record used to document follow-up and treatment of the patient) and the secondary record. A record was deemed secondary if it did not have a problem list and/or had fewer clinical notes than the other record⁷.

We focused on laboratory tests that were ordered in the outpatient clinics and that mandate particular reference from the ordering physician but would not result in a referral for immediate admission to a hospital. We chose the four most common tests that met these definitions: Hemoglobin A1c (HgbA1c) at a level > 8%, Thyroid Stimulating Hormone (TSH) at a level 7μIU/mL or < 0.03 μIU/mL, vitamin B₁₂ < 200 pg/ml and Low Density Lipoprotein (LDL) > 180 mg/dL. These thresholds were based on the reference levels at our institution and the published literature.⁸ We reviewed all the abnormal tests recorded in duplicate records between January 1^st 2008 and December 31^st 2011 (n=452). We only included cases where the duplicate record existed at the time laboratory tests were ordered. We then randomly selected 452 non-duplicate records with similar abnormalities. Records were matched on a case by case basis according to laboratory type and ordering physician. We included all abnormal laboratory results in the records apart from multiple tests drawn on the same day (in such a case one test was selected randomly). We matched records that had multiple abnormal test results with non-duplicate records with a similar number of abnormal tests. Laboratory tests that were not assigned to a specific physician were matched after a review of the record according to the most likely physician who had ordered the test (based on the most recent clinical encounter before the laboratory was drawn). In such cases we matched laboratory tests that were also marked as unassigned from the same physician or the same specialty (i.e., internal medicine, family practice, endocrinology, etc.)

A single reviewer [E.J.], an Internal Medicine specialist reviewed all the records. A laboratory was defined as missing if it was not referenced explicitly in the record within 14 days (i.e., mentioned in a clinical note or a physician comment on a laboratory report, noted in an internal communication, in a documented communication with the patient or implied by a prescription for an associated medication). We also determined whether the laboratory test was performed for a known diagnosis (e.g., HgbA1c in a patient with diabetes) or as a routine screening test (e.g., TSH in a patient without a known thyroid disease). Finally we recorded the content of the secondary records (i.e., presence of clinical notes, laboratory results, imaging reports etc.).

Statistical analysis

Data was analyzed using SPSS (SPSS ver. 20, IBM Inc., Chicago IL). We used the Pearson Chi² and the Student T test to compare categorical and numerical variables between the duplicate and non-duplicate groups (other than the outcome variable that was dependent on the matching). We used general estimating equations (GEE) for binary outcomes with a logit link to model the association between the type of record (duplicate/non-duplicate) and the rate of missed laboratory results. We chose the GEE method in order to account for matching and to control for the effect of other explanatory variables such as demographics and presence of an established diagnosis.⁹ This study was approved by the Committee for the Protection of Human Subjects (Institutional Review Board) at the University of Texas Health Science Center at Houston.

Results

We reviewed a total of 904 abnormal laboratory results. Of these, 452 laboratory results originated in 297 duplicate records and 452 laboratory results originated in 369 non-duplicate records. There were more records in the non-duplicate group because we matched cases by the ordering physician, and a single record could have had multiple laboratory tests ordered by different physicians. Laboratory tests were associated with 102 physicians, each responsible for an average of 4 ± 7 tests (mean ± SD). In total we reviewed 304 cases (33%) of elevated TSH, 132 cases (15%) of low TSH, 278 cases (31%) of elevated HgbA1c, 166 cases (18%) of high LDL and 24 cases of low B12 (3%). There was a lower rate of Caucasian ethnicity in the duplicate group (32.5% vs. 42.3%, p=0.02). There were no other significant differences between the groups with respect to demographics (α = 0.05) (table 1).

Table 1:

Demographics and average laboratory values

	Duplicate	Non-Duplicate	p Value
Ethnicity
Caucasian	147 (32.5%)	191 (42.3%)	0.02
African American	105 (23.2%)	94 (20.8%)	0.38
Latin American	47 (10.4%)	37 (8.2%)	0.25
Asian/Pacific islander	5 (1.1%)	2 (0.4%)	0.25
Unknown	148 (32.8%)	128 (28.3%)	0.15
Sex/Age
Female	298 (65.9%)	285 (63.1%)	0.37
Age	49.4±20	49.7±22	0.85
Laboratory values
Low TSH (n=66)	0.01±0.006	0.01±0.005	0.57
High TSH (n=152)	23.5±33.6	22.4±22.6	0.75
HgbA1c (n=139)	9.8±1.8	10.0±1.8	0.36
Vit B12 (n=12)	149±38	163±24	0.3
LDL (n=83)	200±25	207±26	0.12

Most secondary records had very little data; usually they contained only a single laboratory result or a single scanned clinical note. Abnormal laboratory results were reported in the secondary record in only 18% of the cases (table 2).

Table 2:

Frequency of various data objects within secondary records

Data type	Number of records (%)
Clinical notes	72 (16%)
Problem list	65 (14%)
Laboratory results	231 (51%)
Imaging report	36 (8%)
Medications	108 (24%)
Allergy	20 (4%)
Abnormal laboratory in secondary record	80 (18%)

Pediatric patients had a considerably lower rate of missed laboratory results (6% vs. 35%, p<0.001). There was no difference in the rate of missed results between other specialties or between different laboratory tests. After controlling for age, gender, and known prior diagnosis, duplicate records were associated with a higher rate of missed laboratory results (36% vs. 28%, OR=1.44 95% CI 1.1–1.9). Additional factors that conveyed a higher risk of failing to note an abnormal laboratory result were: tests performed as a routine screening (in the absence of an associated prior diagnosis) (OR=2.24, 95% CI 1.4–3.5) and older age (OR=1.15 for every decade, 95% CI 1.01–1.2). Lab results reported in the secondary record (n=80) were missed in 52.4% of the cases, but there was no significant difference from the matched non-duplicate records (52.4% vs. 47.6%, p=0.7). In a repeated analysis after excluding these cases, duplicate records were still associated with an increased risk of missing a laboratory result (OR=1.51, 95% CI 1.1–2.0).

Discussion

The introduction of new information systems into healthcare may introduce new forms of patient harm.¹⁰ Duplicate records pose such a threat. We found that duplicate records were associated with a higher rate of missed laboratory results (36% vs. 28%, OR=1.44 95% CI 1.1–1.9).

Laboratory results were not missed merely as a consequence of being reported into the secondary record. Over 80% of the results were reported into the primary record, and the majority of the secondary records had very little data, and would not have been easily mistaken for an active record. Why, then, were duplicate records associated with a higher missing rate? One possible explanation is that, duplicate records disrupt the workflow. For example the extra work associated with the need to locate the correct record may discourage physicians from following up on results or from noting the abnormality.

We found the rate of missed laboratory in non-duplicate records to be 28%. This rate is similar to previous studies of missed laboratory results in ambulatory settings. ⁵ However, our rate is considerably higher than the 6.8% missed laboratory rate reported by Singh et. al. for a population of ambulatory patients from the same area,¹¹ and the 18% missed rate reported by Edelman et. al. for HgbA1c.¹² As opposed to our study, Singh et. al. reviewed extremely abnormal laboratory results (e.g., HgbA1c > 15%) that were accompanied by a computerized alert, while, Edelman et. al. used a very lenient measure to define a test as acknowledged (i.e., the presence of an ICD-9 code for diabetes). Interestingly, in Edelman’s study, of the 304 patients with abnormal glucose levels and without a diagnosis of diabetes, 258 did not have any mention of the abnormal laboratory result.¹² This is similar to our finding that laboratory tests ordered in patients without an associated prior diagnosis were more likely to be missed. A similar finding was also noted by Singh et. al.¹¹

A second interesting point is that duplicate records had a significantly lower representation of Caucasian patients. This phenomenon has been reported previously by DuVall et al, who hypothesized that patients from minority groups are at a higher risk of duplicate records due to names that are similar to each other and relatively unfamiliar to hospital or IT staff.¹³

Our study has several limitations. First, we have evaluated a single EHR in a single institution. It is therefore not certain our results are generalizable to other institutions with other clinical systems and administrative processes. In our data, abnormal laboratory results were reported in the secondary record in less than 20% of the cases. A different system might have more laboratory reports directed to the secondary records, resulting in a higher rate of missed laboratory tests. Conversely, another institution may have more stringent protocols for the reporting of abnormal laboratory results, thus reducing the number of missed results. For example, in our study pediatric patients had fewer missed laboratory results.

Another limitation is that the review of the records was done by a single reviewer, not blinded to the objectives of the study. To mitigate possible biases we chose to evaluate objective measures (i.e., whether the laboratory was referenced in the physician’s notes) and refrained from subjective evaluation (e.g., whether the physician took an appropriate action following the laboratory report). This may have resulted in an over/under estimation of the rate of missed laboratory tests. For example, a physician may have noticed the laboratory result and acted upon it, but failed to note it in the patient record. Nonetheless, we regarded such cases as missed because failure to note abnormal results impede the quality of care (e.g., another physician may not know that the abnormal result was noted) and could expose the clinician to legal liability as there will be no written record of the actions taken. On the other hand, a physician may have noted an abnormal result but did not change the treatment. Such cases, may in practice be missed results, but were not considered as such in this study. Regardless, these limitations should not affect the comparison between the matched groups. Further, in our study laboratory results were missed in 28% of the non-duplicate records. This rate correlates with previous published reports.⁵

A third limitation is that we limited our study to cases of duplicates where both records contained clinical data. Such records represent only 6% of the duplicate records in our organization. It is possible that this caused us to overestimate the number of missed laboratory tests within duplicates, yet even if we are to limit our conclusions to duplicate records with clinical data these represent thousands of records. Finally, in our organization duplicate records are being actively reconciled by the clinical IT department. This, also, may have resulted in an underestimation of the rate of missed laboratory results.

The waste associated with missed results includes costs for tests that are never reviewed, the morbidity and mortality of the resulting treatment delay, and needless duplication of service.¹⁴ However, the potential implications of duplicate records on the quality and safety of healthcare extends beyond missed laboratory results. Many of the secondary records held other types of clinical data (clinical notes, reports of imaging studies, documentation of medication and allergies etc.) in addition to laboratory results. These too could have significant clinical effects. Incomplete documentation of patient history may cause suboptimal care. A missed allergy could end in the prescription of a dangerous drug, and a missed prescription could result in contradictory therapy. In addition the extra effort associated with navigating the duplicate records may interfere with the work of physician and cause errors.¹⁰ Finally, the presence of a considerable number of duplicates impedes efforts to generate new knowledge, as there is an overrepresentation of certain patients with fragmented data within the database. For example we might want to run a query to evaluate how many of our patients are adequately screened for diabetes – this will prove to be a difficult task if the data of many patients are not stored within their primary record. Fragmented data also impede efforts to implement systems for monitoring and decision support (e.g., a system that alerts when two prescribed drugs interact will not function if the medications are listed in separate records).

Conclusions

Duplicate records are associated with a higher risk of missing important laboratory results when compared with non-duplicated records. This risk is not merely the result of misallocation of laboratory reports, and may result from a change in clinical workflow. To the best of our knowledge this is the first attempt to accurately quantify the potential harm associated with duplicate records. Further research is needed to establish other facets of the harm associated with these records and ways to mitigate it.

Figure 1:

Study methodology