HIM 690 Module 4 Final Project Milestone Two Example

Final Project Milestone Two: What We Found. Descriptive Results of the Cimarron Heights Flag Accuracy Capstone

[Student Name]

Southern New Hampshire University

HIM 690: Health Information Management Capstone

Final Project Milestone Two

[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.

Final Project Milestone Two: What We Found. Descriptive Results of the Cimarron Heights Flag Accuracy Capstone

This milestone reports the descriptive findings of the Cimarron Heights capstone: how the sample looks, how often coded admission flags agreed with the blinded second reading, which way the disagreements ran and how often a structured admission assessment was documented. Inferential results, the logistic model and the interview themes, follow in Milestone Three. All figures come from the final cleaned data set.

Sample

Of 318 records drawn, one could not be analyzed because transfer documents were never scanned, leaving 317. The final sample contains 80 records each for pressure wounds of stage 2 or deeper, catheter-related urinary infections and venous blood clots, and 77 for fall-related injuries, all records available in that stratum after exclusions. Patients had a median age of 71, 54% were women and 23% were admitted from a nursing home or other facility, 62% through the emergency department and 15% directly. Records came from nine inpatient units, including two intensive care units.

Reliability of the Gold Standard

Across all three batches, 32 records were abstracted twice. The two coders agreed on 29 of them, a kappa of 0.85 for the whole study, with batch values of 0.86, 0.88 and 0.82. Every batch stayed above the planned floor of 0.80 and within the range McHugh (2012) describes as strong. The gold standard is therefore reliable enough to judge the coded flags.

Agreement

Table 1 shows agreement between coded flags and the second reading for each complication. Overall, 242 of 317 coded flags agreed, 76.3%. Because the sample took equal numbers from each complication while the hospital's year contained many more blood clot cases than pressure wounds, the overall figure was also weighted by each complication's share of the 557 eligible discharges, giving an estimated agreement of about 77% for the hospital's full year. The figure is close to the 74% agreement Goldman et al. (2011) found across California hospitals, which suggests Cimarron Heights is not unusual overall.

Table 1. Agreement Between Coded Flags and Re-Abstraction by Complication

Note. Confidence intervals use the Wilson method. Weighted overall agreement for the hospital year is about 77%.

Pressure Wounds Stand Apart

The overall figure hides a large difference. Pressure wound flags agreed with the chart in only 61% of records, and the confidence interval for pressure wounds does not overlap with those for urinary infections or blood clots. In practical terms, about two of every five pressure wound flags in the sample were wrong.

Disagreement by Admission Source and Unit

Two further descriptive patterns help frame the analysis. Disagreement was more common for patients admitted from nursing homes and other facilities, 27 of 73 records or 37%, than for patients admitted through the emergency department, 39 of 197 or 20%, or directly, 9 of 47 or 19%. Among the nine units, disagreement ranged from 11% on the unit with the required admission template to 36% on the busiest medical unit, which also had the lowest share of documented assessments. Intensive care records disagreed less often than general unit records, 15% compared with 25%, perhaps because intensive care nurses document skin findings on a fixed schedule.

These differences overlap with one another and with the assessment variable. Facility admissions arrive disproportionately on the busiest medical unit, and that unit documents fewest assessments. Descriptive tables cannot separate these influences, which is exactly why the next milestone uses a model that considers them together. They are reported here so that readers can see the raw patterns before any adjustment and judge whether the model's results make sense.

Direction of Disagreement

Of 75 disagreements, 44 were over-reports, in which the coded flag said the condition was present on admission and the second reading found it was not, and 31 were under-reports. Across all complications, that split could have occurred by chance; an exact binomial test of 44 versus 31 against a fifty-fifty expectation returned p of roughly 0.17, two-sided. For pressure wounds, however, 26 of 31 disagreements were over-reports, and the same test gave p below 0.001. The other three complications showed no consistent direction. Bahl et al. (2008) showed that treating conditions as present on admission lowers reported hospital-acquired rates, so the pressure wound pattern means the hospital's reported rate of pressure wounds acquired in its care is likely understated.

Payment Effect Estimate

As planned, a separate estimate examined payment. Of the 44 over-reports, 19 involved conditions on the federal list of hospital-acquired conditions for which payment is not increased when the condition develops during the stay. In each of these, a flag of present on admission may have allowed a higher payment than the record supports. The writer recalculated the 19 claims with the corrected flag using the hospital's grouping software; 7 would have been paid less, by a total of about $41,000. These claims have been reported to the compliance office as agreed in the charter. The estimate is kept separate from the accuracy findings and is not extrapolated to the full year, because the sample was not designed for that purpose.

Admission Assessments

Charts held a completed structured skin and risk assessment from the first 24 hours in 222 of 317 cases, 70%. The share ranged from 61% to 92% across units, with the highest on the unit whose admission template requires the assessment. Disagreement was more common when the assessment was missing, 41 of 95 records compared with 34 of 222, but whether that difference holds after accounting for complication, admission source and unit is a question for the model in Milestone Three.

Summary

The flags agree with the chart about three times in four overall, a typical result. Pressure wounds are the exception: flags are wrong in two of five records and lean strongly toward calling hospital-acquired wounds pre-existing. Missing admission assessments are common and appear linked to errors. These findings set up the analysis to come.

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

McHugh, M. L. (2012). Interrater reliability: The kappa statistic. Biochemia Medica, 22(3), 276-282. https://doi.org/10.11613/BM.2012.031