HIM 675 Module 2 Literature Review Short Paper Example

Trusting the Flag: A Review of Research on Present-on-Admission Reporting Accuracy

[Student Name]

Southern New Hampshire University

HIM 675: Research Methods and Evaluation

Module Two Short Paper

[Instructor Name]

[Date]

The organization, setting and figures below are a composite written as a model document. No real employer, client, colleague or patient is described.

Trusting the Flag: A Review of Research on Present-on-Admission Reporting Accuracy

Since 2008, Medicare has required hospitals to report, for each diagnosis on an inpatient claim, whether the condition was present when the patient was admitted. The flag decides whether a pressure injury, an infection or a fall injury is counted as something the hospital caused, which affects quality reports and, for selected conditions, payment. A hospital that plans to study its own flags should first know what earlier research has found. This review covers three themes in that literature: the value of the flag for measuring hospital performance, the accuracy of the flag compared with the medical record and the effect of flag errors on safety indicators. It ends with the gaps a local study could address. Articles came from PubMed and CINAHL queries joining present on admission to accuracy, validity and reabstraction, then followed the reference lists of useful articles.

Why the Flag Matters for Measurement

Before the flag existed, claims could not tell a comorbidity from a complication. Pine et al. (2007) tested how much risk-adjustment equations for inpatient mortality improved as more clinical detail was added to administrative data from 188 Pennsylvania hospitals. Adding present-on-admission codes, along with laboratory values available on admission, substantially improved how well the models separated patients who died from those who survived. The study made a strong case that the flag could make administrative data far more useful for comparing hospitals, provided the flag itself was reported accurately. That proviso is where the rest of the literature begins.

How Accurate the Flag Is

The most direct test of accuracy is to re-abstract records blind and compare the result with the coded flag. Goldman et al. (2011) did this for 1,059 records from 48 California hospitals, choosing secondary diagnoses that strongly predict mortality among patients with heart attack, heart failure, pneumonia or coronary angioplasty. The coded flag agreed with the gold standard in 74.3% of records. Errors ran both ways: 13.7% of records over-reported conditions as present on admission and 11.9% under-reported them. Accuracy also varied by hospital type. For-profit hospitals were more likely to over-report, which would make complications look like pre-existing conditions, and teaching hospitals were more likely to under-report, which would make pre-existing conditions look like complications.

Two features of this study matter for a hospital planning its own work. First, it measured over- and under-reporting separately, which is more informative than a single accuracy figure because each error has a different consequence. Second, its conditions were mortality predictors in medical and cardiac patients, not the hospital-acquired conditions that drive safety reporting, such as pressure injuries and catheter infections. Its findings may not transfer directly to those conditions, which are documented by nurses as much as physicians.

What Flag Errors Do to Safety Rates

Bahl et al. (2008) approached accuracy from the other side, asking how much patient safety indicator rates changed when conditions present on admission were removed. Using one academic health system's discharges, they found that rates for all but one of 13 indicators fell when the flag was applied, and that the drop was statistically significant for five, including decubitus ulcer and selected infections due to medical care. A comparison with an earlier chart review at the same system gave consistent results. The authors concluded that indicators which cannot separate pre-existing conditions from complications overstate harm. Read alongside Goldman et al. (2011), the implication is that flag accuracy determines whether safety rates are meaningful at all: a falsely positive flag hides harm, while a falsely negative one invents it.

Data Quality as a Framework

These studies sit within a wider literature on electronic health record data quality. Weiskopf and Weng (2013) reviewed methods for assessing record data quality and identified recurring dimensions, such as missing values, wrong values, clashes between two sources, values that make no clinical sense and values that are out of date, along with common assessment methods such as comparison with a gold standard and agreement between elements. Present-on-admission accuracy is a problem of correctness and of concordance between the coded claim and the clinical notes. Their framework offers a vocabulary for a local study and suggests that more than one method, for example gold-standard comparison plus checks for internal consistency, gives a fuller picture than one alone.

Comparing the Methods

The studies use three different methods, and the choice shapes what each can claim. Pine et al. (2007) judged the flag by what it added to a statistical model, an indirect test that shows usefulness but not correctness. Goldman et al. (2011) used blind re-abstraction by trained reviewers as a gold standard, the most direct test of correctness, though a gold standard built from the same record cannot catch conditions that were never documented anywhere. Bahl et al. (2008) compared indicator rates with and without the flag and checked them against an earlier chart review, which shows consequences but depends on that earlier review's quality. A local study gains most by combining the direct method, blind re-abstraction with a reliability check between reviewers, with a measure of consequence, such as how the hospital's reported rates would change if the corrected flags were used.

Gaps in the Literature

The literature leaves at least three gaps. First, the large accuracy studies are now more than a decade old and predate widespread electronic nursing documentation, which may have improved or worsened accuracy. Second, they focus on mortality predictors rather than the nursing-sensitive conditions that drive hospital-acquired condition reporting. Third, they describe how often flags are wrong but say little about why, for instance whether a missing admission skin assessment leaves coders without evidence. A single-hospital study that re-abstracts flags for pressure injuries, catheter infections, blood clots and fall injuries, separates over- and under-reporting and tests whether documentation of an admission assessment predicts accuracy would speak to all three gaps, while recognizing that one hospital's results cannot be generalized widely.

Conclusion

Research shows that present-on-admission flags can make administrative data much more useful, that they agree with the chart only about three times in four and that errors meaningfully change reported safety. What it does not show is how accurate the flags are today for nursing-sensitive conditions or what documentation practices drive the errors. That is the space the proposed study at Cimarron Heights will occupy.

References

Bahl, V., Thompson, M. A., Kau, T.-Y., Hu, H. M., & Campbell, D. A. (2008). Do the AHRQ patient safety indicators flag conditions that are present at the time of hospital admission? Medical Care, 46(5), 516-522. https://doi.org/10.1097/MLR.0b013e31815f537f

Goldman, L. E., Chu, P. W., Osmond, D., & Bindman, A. (2011). The accuracy of present-on-admission reporting in administrative data. Health Services Research, 46(6pt1), 1946-1962. https://doi.org/10.1111/j.1475-6773.2011.01300.x

Pine, M., Jordan, H. S., Elixhauser, A., Fry, D. E., Hoaglin, D. C., Jones, B., Meimban, R., Warner, D., & Gonzales, J. (2007). Enhancement of claims data to improve risk adjustment of hospital mortality. JAMA, 297(1), 71-76. https://doi.org/10.1001/jama.297.1.71

Weiskopf, N. G., & Weng, C. (2013). Methods and dimensions of electronic health record data quality assessment: Enabling reuse for clinical research. Journal of the American Medical Informatics Association, 20(1), 144-151. https://doi.org/10.1136/amiajnl-2011-000681