first case study was done on a 76-year-oldmale who was admitted to the hospital with ammonia. The physician who was creating the physicaland history on this patient noticed that the patient was admitted to thehospital about six months ago with the same issue. So, when creating theHistory and Physical (H&P) for this visit the provider copied someinformation from the prior visit and pasted it into the new note. One of the lines that were captured was thatthe patient had hypertension, diabetes type 2, and a urinary tract infection(UTI).
As the patient was treatedthroughout the visit and was eventually discharged the coding departed receivedthe documentation for this visit with the new diagnosis of pneumonia,hypertension, diabetes type 2 and a urinary tract infection (UTI). The coding outcome was of this account was aDRG of 194 and a reimbursement of $4,069 dollars. The issue was that theurinary tract infection (UTI) from the prior visit was not an issue on theformer visit and coding for the diagnosis resulted in a complications or comorbidity(CC) however removal of the UTI changes the DRG to 195 simple pneumonia andpleurisy without a complication (CC) and the reimbursement dropped by over athousand dollars to $2, 959. This is anexample where leaving in information in this case of a prior diagnosis isn’timplacable to the current visit had an impact on the coding and resulted inoverpayment. CASE STUDY 2 The second case study was done on a 63 yearold woman admitted in the morning to a hospital for congestive heart failure (CHF. Upon admission some additional lab test weredone one of which showed that the patient had a potassium level of 5.
1 and thenormal range for potassium is from 3.7 to 5.2. In creating the documentation for this visit specifically the historyand physical note the admitting physician notices that the patient had someprior admissions due to her congestive heart failure (CHF), copied some contextfrom the prior visit notes, pasted it into the current history and physical(H) in order to speed up the creation of the note. In doing so one of the items that wascaptured was the patient’s potassium level in admission for a prior visit.
On the prior visit the patient’s potassium levelwas to 2.9, which is below the normalrange and in copying that information forward and not editing it made it appearas if the patients potassium level in the current visit was 2.9. Thus, moving forward to the evening shift newhospital staff arrive to cover this patient. The evening shift hospitalist reads the history & physical(H), for the new admission and notes that the potassium level is low at2.9.
This hospitalist then orderedpotassium supplements and a recheck for the potassium levels for the next day. The issue in this case study is the mix matchpotassium levels because the patient was actually within the normal rangehowever the documentation made it appear as if the patients potassium level waslow. Two issues arise from theseencounters and the most important one is an adverse event risk becauseadministering potassium supplements to a patient with normal potassium levelscan have a devastating impact on the patient. The second is avoidable cost ordering a re-check of the potassium levelis an additional cost that could have been prevented. So, here there is a second instance ofcopy-paste being used in the electronic health record (EHR) and in manyinstances no issues would have occurred but now we have just one bit ofinformation that proves that copy forward can have some adverse event risks forthis patients encounter. SPELLING ERRORS IN CLINICALDOCUMENTATION Accurate medicaldocumentation is critical for safe patient care and effective inter-providercommunication.
Medical documentationerrors can lead to some causes in injury and or even patient deaths, and isalso important for the care correlation between providers. Studies show that about 5 million errors peryear are tied to wrong documentation involving drugs that look and sound alikesuch as; Altenol vs. Alendol or Lyrica vs. Lamictal. These spelling errors can happen due tonon-word errors such as; Humulog for Humalog and are commonly due to free-textentries (typed notes) or real-word errors (words spelled correctly but arecontextually wrong, such as; (there for or their) Speech Recognition (SR)generated text. Studies used sourcesincluding a couple of standard medical terminology such as; UMLS, SNOMED, CT,RxNorm and Cetera.
The error correction was based on Shannon’s noisy channelmodel; specifically using the Daneray-Levenshtein edit distance betweenmisspellings and the suggestions, both in terms of their orthography andphonetics. To evaluate and compare theirsystems performance, they used ASpell Default setting as their baseline. It isan open, free software spell checker, which helped to show a significantimprovement in the spell check regarding precision, and accuracy. Although, speech recognition technology hasbeen widely used in medical practices, the quality and accuracy of clinicaldocuments hasn’t been thoroughly studied or reported.
The limited scope and sample size, for thestudy used a discharge summary and a progress note, and another involved twophysicians dictating 47 emergency department