A COPD risk score was calculated based on the COPD-PS screener algorithm.7 The COPD-PS screener uses an algorithm consisting of age, lifetime cigarette use and smoking symptoms to identify patients who were offered spirometry can be screened for COPD. A total of 30 percent (198 patients) had a healthy lifestyle and no indication for further examinations or preventive services.
Questionnaire
Under-coding of secondary conditions in coded hospital
Mingkai Peng, Danielle A Southern, Tyler Williamson and Hude Quan
Introduction
The Canadian coding standard is maintained and developed by the Canadian Institute for Health Information (CIHI) based on ICD-10 developed by the World Health Organization (WHO). Based on validated algorithms, we identified the conditions listed in the Charlson and Elixhauser comorbidity indices.12 The Charlson and Elixhauser comorbidities include 31 conditions and are two commonly used instruments for risk adjustment analyses.
Statistical analysis
A detailed description of the card review process can be found in our previous publication.4.
Results
Discussion
The sensitivity for the four conditions increased as the total number of diagnosis codes in the record increased. The number of research studies based on administrative health data has increased dramatically in recent years.
Limitations
Coding validity could be dramatically improved if all conditions were coded, regardless of whether those conditions are clinically implicated in hospitalization. It is suggested that certain important chronic or modifiable conditions, such as hypertension and diabetes, should be coded as long as they are documented in the chart.
Conclusion
Assessing the validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique double-coded database. A modification of the Elixhauser comorbidity measures to a point system for hospital death using administrative data.
Health Information Exchange
What do patients want?
Laura N Medford-Davis and Lawrence Chang
Rhodes
Of these, our primary outcome question was answered as to whether they were willing to share their data in the HIE and therefore included in the analysis. Of the patients who reported having outpatient providers in the same hospital system as the ED (N expected that their EDs would be able to see their records from those outpatients.
Diagnosis of adverse events after hysterectomy with postoperative
A pilot study
Donna T Gilmour, Norman J MacDonald and Steven Dukeshire
Barbara Whynot
Barry Sanders
John Thiel
Sony Singh and Craig Campbell
Krisztina Bajzak
Gordon Flowerdew
Adverse events and their consequences after elective hysterectomy can be significant and costly for patients, surgeons, the health care system, and society. These less severe adverse events may be managed inefficiently and ineffectively if patients present to inappropriate health care settings for assessment and treatment. Since adverse events after hysterectomy that occur at home are relatively common and potentially serious, we also need to ensure that these resources provide accurate, timely, and reassuring information tailored to their symptoms and type of surgery.
Currently, there are no formal screening mechanisms in place to rapidly, routinely, and comprehensively screen patients for specific symptoms of adverse events after hysterectomy and advise them when and where to seek appropriate care. 282 Health Informatics Journal 23(4) women who experienced adverse events during the study and how using the app affected their care and outcomes.
Materials and methods
The incidence of adverse events was determined by medical research staff conducting a medical chart review and contacting each participant to capture adverse events not documented in their medical chart. Because the sample sizes are too small to draw definitive comparisons between groups and the fact that the pilot cohort largely followed the same methodology with minor revisions after the feasibility phase, the experiences of all 31 women are reviewed as one cohort in this article.
Results Demographics
We were able to contact and interview 8 of the 11 women who experienced an adverse event. Six of the 31 women (19%) did not use the web applications regularly (completed <50% of the sessions). Two of the women interviewed also suffered an adverse event (Table 2, first two entries).
Women who experienced an adverse event tended to give higher scores for the questions in the questionnaire to assess the helpfulness and usability of the web applications (Table 3). If they gave a positive answer, they were given information about the possible meaning of the symptom and where to go for assessment.
Preserving medical correctness, readability and consistency in
Kostas Pantazos * , Soren Lauesen and Soren Lippert
Consistency: The patient identification data must be consistent with the medical picture, for example an age that resembles the age of the real person. If the real patient's name is Peter, but his anonymized name is Jens, Peter should also be replaced by Jens in the doctor's free text notes. These approaches de-identify database records (e.g., pathology papers) that are unrelated to other records.
In this way, patient B received patient C's first name, patient D's last name, patient E's street name, and so on. To our surprise, 3-4 percent of the words in the free text fields were potential patient identifiers.
Related work
In 43,119 cases, the data could not be securely de-identified without manual intervention, so we made sure the program deleted these patient records. The Swedish deidentification system was developed by Kokkinakis and Thurin7 using named entity recognition. This phenomenon was also observed and confirmed by Grouin et al.9, who adapted the de-identification system from English to French and obtained poor results.
Meystre et al.23 reviewed recent de-identification algorithms and found that the majority of algorithms focus on de-identifying structured data and not free text. However, in accordance with Dalianis and Velupillai24 and Hanauer et al.12, there is enormously valuable information in the free text.
Quality factors
This approach de-identified 200 Swedish discharge letters with a precision of 96.97 percent, recall of 89.35 percent, and f-measure of 93. If we replace a medical term similar to a personal name with the new personal name, the health record will look strange . In many cases, a clinician can guess what the medical term was and thus learn the original name of all patients who have this new name.
If birth dates are transformed in such a way that the patient is given a completely different age, this will not match the patient's diagnosis pattern.
Solution
If he knows the substitution rules, he now knows that everyone in the database named C is actually B. If the word is actually a person's name, the clinician can see it from the context. If he knows the substitution rule, he now knows that the referred person is really called Aaron, even though that is not his name in the de-identified database.
If there are only a few Aarons in real life, he might guess who it is. We've lost a bit of consistency, but such data could still exist in the database.
The database and the mapping tables
Many names did not appear in the patient chart at all, but may appear in the free text notes. Rare names (frequency < 200) were randomly replaced by another name in the frequent part of the list. With this, we also took care of names that did not even appear in the patient table.
For names that appear in the patient table, we extracted the gender from the resuscitation number. Just because a name appears in the dictionary does not mean it has another meaning that may appear in health records.
Applying the mappings
As another example, it is common for a city, hospital, clinic or road name to be used as a first or last name. So the medical specialist on our team (Lippert) scrutinized the list and came up with ambiguous names from 1952. So we used the website "Who named it"30 and extracted 3246 medical eponymous names from it.
If the number is in the postcode table and next to the city name, replace it with the new postcode. This can be a measured value, a laboratory test number (eight digits), a house number, etc.).
Evaluation of the quality factors Anonymity, readability and medical correctness
Furthermore, we are not allowed to move our original data outside the company where it is hosted. When a person's name is de-identified in the structured patient chart, it is important that the same name is de-identified in the same way in the remaining patient records and linked patient records, also for free text. field. They are not part of the database, but the fields in the database contain the file names.
Szarvas G, Farkas R and Busa-Fekete R. Modern anonymization of medical records using an iterative machine learning framework. Changing the function of the electronic medical record: creating a de-identified database for clinical researchers and educators.
The psychosocial effect of web-based information in
Lene Bastrup Jørgensen
Lone Ramer Mikkelsen
Bodil Bjørnshave Noe
Martin Vesterby
Maria Uhd
Bengt Fridlund
Due to a significant proportion of patients not using the platform in the E-THAP group, a subgroup analysis was performed to stratify users (those who accessed) versus non-users (those who did not access) of the platform to compare. In contrast, a positive effect was identified by Tou et al.17. No study dealt with orthopedic rapid surgical programs, with measurement times also varying between studies. The validity and reliability of using the VAS anxiety score to assess pre-operative anxiety, and as a predictor for a patient's mental state while participating in a rapid program, needs further study.
This study documented improved psychosocial effects of rapid THAP without the additional effect of online and animated information. Identification of utilitarian motivational factors may be necessary to improve our understanding of the psychosocial benefits of E-THAP and allow us to target rapid THAPs that involve psychosocial challenges.
Extracting and analyzing ejection fraction values from electronic
Fagen Xie and Chengyi Zheng
Albert Yuh-Jer Shen
Wansu Chen
Finally, there are cases where the report contained conflicting information about EF values or textual descriptions suggesting LV systolic functions. The procedure for extracting EF values/text descriptions from ECHO reports is outlined below and also shown in Figure 1. As previously mentioned, some ECHO reports include the patient's previous EF values or text descriptions indicative of LV systolic functions.
If multiple EC values or text descriptions were found in a report, the final value is determined using the following rules. If multiple EC values or text descriptions had the same priority, the one indicating the worst EC function was selected.
Discussions
Sensitivity was defined as the number of reports in which EF values or text descriptions were correctly retrieved by the automated algorithm, divided by the total number of reports in which EF values or text descriptions were retrieved by the cardiologist. PPV was defined as the number of reports in which EF values or text descriptions were correctly extracted by the automated algorithm, divided by the total number of EF values or text descriptions extracted by the automated algorithm. The numerical EF values or text descriptions were successfully retrieved from a total of 621,856 ECHO reports between 1995 and 2011 through the developed automated algorithm.
Despite these limitations, our study developed an automated algorithm to extract EF values from the ECHO reports in a systematic and automated manner. PPV = number of EF correctly extracted by the computer algorithm/total of the number of EF correctly extracted by the computer algorithm.
