Clinical quality measures in an electronic format have existed for quite some time as a major part of the federal Electronic Health Record Incentive Programs, more commonly known as meaningful use. eCQMs had the potential of extracting the precise data pieces needed to record a certain measure of medical excellence without the necessity of human assistance.

The purpose of the first group of eCQMs was to document treatment and services given to patients to see if quality care was given based on scientific standards.

Ten years have passed since the concept of electronic Clinical Quality Measures (eCQMs) was first introduced, yet vendors of Electronic Health Records (EHRs) and healthcare providers alike still have difficulty capturing the existing eCQMs precisely in a way that accurately showcases the quality of the care provided.

Joseph M. Kunisch, who is a registered nurse with Board Certification and the enterprise director of clinical quality informatics, regulatory performance at Memorial Hermann Health System, has a great deal of expertise in electronic Clinical Quality Measures. At a recent HIMSS20 Digital event, he spoke about the difficulties of obtaining correct eCQM data from a clinical standpoint.

A Computer-Coded Algorithm

An eCQM is a computer-coded program that identifies specific data which defines the group being examined and the data that will be used to assess the standard of care. This criteria is based solely on individual data which contains an associated numerical computer code.

The original technique used to pull out quality measures was denominated “chart abstracted.” This process entailed a nurse (or a similar healthcare professional) going through a chart, responding to questions. The eCQM automatically extracts, without human review.

In the chart abstracted approach, one would need to read and comprehend hundreds of pages of intricate directions. eCQMs employ a computer-generated algorithm employing a particular computer language.

Kunisch noted that the most difficult difference between the two was that during chart abstraction, the individual was free to use information from all possible sources, containing both written documents and scanned images.

eCQMs require data from various information systems to be in a coded, distinct format rather than using the data raw. This significantly restricts the places from which one can access the data contained in an electronic health record.

In the transition from chart abstracted quality measures, the EHR basically takes over the job of the human abstractor; thus, IT personnel must comprehend how the computer now extracts that data. Unlike someone who is able to give an explanation of the rationale behind the information they found, an EHR system that is automated is unable to do so.

Disparate Information Systems Challenge

Kunisch mentioned that a major difficulty exists when health care providers utilize dissimilar information systems for eCQM reporting.

He noted that they had a different system for radiologic diagnostics, where the radiologist would vocalize their results. This report was transmitted to the electronic health record system in the form of an image with text.

Once more, it would be simple for a human expert to read through this report and rule out the patient due to the fact that it was apparent the patient already had a blood clot condition. It is not possible to use the text in eCQMs to exclude a patient, as it cannot be read.

A clinician must input an ICD-10 diagnosis or SNOMED issue within 24 hours of the occurrence of the issue.

Kunisch stated that if the relevant providers are asked to specify a diagnosis when trading the report, it will be cause for celebration due to the efficiency of capturing accurate codes at the appropriate moment.

The team predicted this would happen, but even after running our data analysis we weren’t catching these patients that should have been exempted from the quality measure. This is clear as the written report demonstrates they have had a VTE before they were admitted to hospital.

Value Sets That Support eCQMs

It was discovered that Kunisch and his team had overlooked an issue: data that was limited by the values set. Collections of distinctly coded data that assist the electronic Clinical Quality Measures (eCQMs) are known as value sets. A clinician can use the value set authority to locate all the data sets associated with a VTE for every data element.

It is possible to observe the vocabulary system that is employed when dealing with codes related to diagnosis, as well as the amount of entries present in each set. Why is this number important? One must ensure that they meet the criteria in the electronic health record.

Kunisch mentioned that it is feasible to accurately link specified details from one’s hospital’s electronic health record to that of the vendor’s electronic health record within most electronic health record systems.

One of the issues facing us is that not all the material from the electronic health record providers can be adjusted, limiting the range of information obtainable in such mapping tools. This is a difficulty that arises when using the interface that the clinician has access to.

Picture a hectic ER doctor who is told to input a diagnosis. She inputs VTE into the Electronic Health Record and retrieves an extensive number of choices. This further complicates the situation because the codes are not ICD-10, but rather intelligent medical objects which are easier for healthcare providers to comprehend.

The IMO codes are connected with the ICD-10 codes or SNOMED issues. So what does a typical ER physician do? Kunisch declared that she selects either the first item on the roster or the first one nearest the one she is compatible with.

We uncovered that the preferred choice of our practitioners in the search bar located on the left side was not linked to any of the eCQMs value set, Kunisch shared. It seemed that the request did not go through, and it seemed like the patient ended up getting a Venous Thromboembolism while spent time in the hospital because the essential ICD-10 code had not been put in within the initial 24 hours.

To conclude, while much effort has gone into eCQMs in the last decade, there are still notable issues to be addressed. There are some current initiatives that could help.

Included among the technological solutions are FHIR, a new computer programming language for quality metrics; advances in natural language processing, which is a program that can read and interpret text and transform it into distinct information sets; and true interoperability that all medical providers are striving to accomplish.

Kunisch finished by emphasizing that it is essential that every significant participant is engaged if eCQMs reports are to succeed in the future. It is advantageous for everyone to take part in any capacity to ensure the desired outcome.

Understanding Challenges of Using Routinely Collected Health Data to Address Clinical Care Gaps

A learning health system is the key to accomplishing the five objectives of improving care for patients, safeguarding public health, guaranteeing fairness, increasing efficiency, and creating a better experience for healthcare workers, which will all contribute to further aspirations, such as precision in medical treatments. In order to respond quickly to vital clinical inquiries, it is essential to revamp the structure and information gathering of electronic medical records and health databases.

Alberta, a place in Canada, is highly regarded throughout the world for its capacity to store and obtain health data.

Nevertheless, health service investigators have identified crucial drawbacks to its application. The cause of these restrictions is due to the past implementation of various health information systems in Alberta’s regions, coupled with the formation of administrative health databases for duties that don’t require a clinical aspect, such as remuneration.

The PLP is a provincial initiative that attempts to identify any disparities in clinical practices, construct pertinent data that can be used to take action, and team up with health care providers, patients, community members, and health system partners to further progress successfully.

Below are four instances of PLP initiatives concerning an array of conditions, some being rare, others more common. This allows us to observe the obstacles of making use of regular collected health data to solve actual clinical issues and improve quality.

These four initiatives revealed problems with regard to the collection of data, which would give us useful information to enhance the care for Alberta residents if properly addressed. Our team provides assistance in refining usually collected health information that is important to health systems by dealing with problems of completeness, availability, absence, repetition, and various amounts of capture.

It is critical to make changes in these areas in order to make data more useful for healthcare, health services research, and, eventually, potential applications of artificial intelligence and precision health.

Gaining quick access to clinically relevant data is a vital step to forming a strong learning health system to achieve the five goals. It is essential to have a health data system that quickly provides medically relevant information for evidence-based treatment and raising the quality of care, in order to promote skill review and the development of new ideas that address patient requirements.

The endeavors of our PLP bring about awareness of the four issues that come with attempting to employ routinely obtained health information to accomplish our goals. At the beginning, we realized that not all details gathered during patient visits had an associated entry fields in an administrative database. Consequently, costly and time-consuming primary data collection is necessary to answer crucial clinical questions, making continual observation impractical.

When data fields are accessible, sometimes they may not be included at all or there may not be consistent data across the fields. An example of this is when results from medical assessments or tools used at the bedside are not relayed to records belonging to the management staff.

It was hard to find out what illnesses were common and how many times people visited the doctor since there were incomplete records, multiple databases with the same facts which were difficult to compare, various diagnostic coding methods, and different classifications between databases.

An integral part of this problem was that the inexact diagnosis codes, like ‘follow-up’, did not allow for understanding of what the visit entailed. It was difficult to create queries to acquire all pertinent tests because of the wide range of laboratory diagnostic codes employed for the same exam.

The goal of the PLP is to develop useful medical knowledge and partner up with physicians, personnel, patients, and partners to generate long-standing answers to promote the practice. Generating actionable health information from routinely collected data is difficult when there is significant missing data.

To measure progress, baseline data and Continued tracking is needed to compare progress over time. The work of Burles et al, Clement et al and Edmondson and Reimer have gone into great detail in discussing the pros and cons of administrative and electronic medical record health databases.

The challenge of not being able to process data in an immediate manner is not just found in Canada, with it also being an issue which has been documented in other countries, including the United States. There are common difficulties that arise when dealing with information gathering, quality, and consistency across healthcare services and environments.

Furthermore, there have been numerous issues brought to light when working with digital medical records. These difficulties are likely to be seen all over Canada since a few of the provided databases, for example Discharge Abstract Database and NACRS, are accessible nation-wide. The PLP is continually collaborating with pertinent parties to handle the matters presented.

We understand how important it is to work together with several stakeholders, such as data scientists, medical professionals, and superiors, to discover which clinical data is most influential and how to bring it into use in order to facilitate quality improvement through data-driven decision making

. Better collaboration and taking advantage of the new provincial acute care electronic medical record should help drive this task forward, especially as activities develop around the care cycle.

Future Directions

It is essential to elevate the quality of health care data for existing quality enhancement initiatives, as well as to make use of precision health and artificial intelligence techniques for furthering healthcare in the future.

It is essential for health care systems to have data in order to achieve the purpose of the Federation of Medical Regulatory Authorities in Canada, i.e. for all Canadian practitioners to partake in the improvement of practice quality by means of data-driven methods.

The main objective of these attempts is to foster the formation of an instructional health system and to accomplish progress in the fivefold aim of boosting general health, patients’ care experience, equality, cost effectiveness, and sustainability of healthcare workforce.

The advantages in the distant future resulting from an improved collection of data would more than make up for the costs of the initial investments. It is significant to back up these efforts by managing clinical details in such a manner that does not tax the medical experts and lead to exhaustion for the doctors.

It is critical to tackle the four issues that have been identified if we want to generate a learning health system and enhance healthcare services and health results. We recommend the following:

• To have more clinically important data available in readily extractable formats, we suggest expanding and harmonizing mandatory data submission requirements with increased clinician engagement to ensure data that is captured is clinically meaningful.
• To increase the quality and validity of the data available to assess patient care, we suggest the use of more specific codes and consistent taxonomies across the healthcare system to capture encounter diagnoses; standardization of data entry processes with clear mechanisms of training and maintenance; and, ensuring the flow of clinically important information from bedside instruments, laboratory settings, and diagnostic imaging results to administrative databases in analyzable formats.
• To enhance efficiency and speed of data capture so that upgrading data quality, quantity, and structure is not at the cost of the clinical user, we suggest the incorporation of technologies like natural language processing, cross-platform interoperability, and application of human-centered design for workflow process improvement.
• To promote real-time usability of data, we propose integrating technologies such as natural language processing and artificial intelligence to automate routinized functions to support appropriate real-time clinical decisions and reduce clinician burden.