Connect with us


In Healthcare, We’re Dealing with Another Type of Attachment Disorder



The Council for Affordable Quality Healthcare (CAQH) released its 2020 Index earlier this year, which tracks progress toward automating various administrative processes. As a result of converting manual processes and the data they use to electronic transactions, billions of dollars have been saved, with 2019 savings increasing by $20 billion to $122 billion. Yes, that is $122 billion, an enormous sum for anyone who is not Jeff Bezos.

However, even though transactions such as claim submission are highly automated, with estimates exceeding 90%, certain processes represent a considerable cost in terms of both money and time. Consider the issue of attachments.

Although the overall volume of attachments in medical transactions is rather low (less than 1% of all transactions), they continue to be a significant hassle. While attachments are not an administrative transaction in and of themselves, they are an integral part of various processes, including claim submission, prior authorization, audits, and referrals. And when they are involved, progress frequently comes to a grinding halt.

To fully grasp the issue, it’s necessary to approach it from a data-driven perspective. Most of the time, attachments are medical records such as lab results, progress notes, and physician orders. Even when electronic medical records systems are implemented, which eliminates the need for paper records, most of the data they contain is still what we refer to as “unstructured data.” It is considered unstructured because a large portion of what doctors and other medical personnel input into the system are unlabeled text. This means that the ability of a system to easily convert data contained within it—a diagnosis, for instance—to another format is extremely limited. And when it comes to sharing data stored in record systems, regardless of whether the system is a paper-based colored file system or a modern web-based solution, processes tend to be archaic.

CAQH quantifies attachment automation for two processes: claim submission and prior authorization, and believe it or not, the primary method of sharing these documents is via fax or mail. CAQH estimated in a separate report that over 80% of attachments arrive in this manner. This means that providers frequently print records to share them, converting digital information (even if it is unstructured) to images or paper, making a separate document handling and data entry process necessary on the receiving end to manage them. One regional health plan that took part in the study revealed that it manages the equivalent of twenty full-time employees solely to process attachments.

The cost of managing attachments is high, even with relatively low transaction volumes. Several factors contribute to the persistence of attachment disorders.

Complexity of Data. Even if they are born digital, medical records are not always structured. While electronic medical records systems allow for form-based data entry for common data such as patient data and service date, a large portion of data contained within records is unstructured text, making the process of locating specific information complex and time-consuming.


Inadequate standardization. Unlike other transactions for which the Department of Health and Human Services, of which CMS is a part, has expressed strong support for electronic standards, medical records continue to be a thorn in the side of medical transactions. As discussed previously, a large part of this is due to the nature of information. However, another issue is the diverse agendas of EMR/HER vendors, who are not particularly interested in facilitating data conversion between systems; there is an incentive to promote “vendor lock-in.”

Inability to begin !00% Digital. Medical records, even those that are several years old, are a valuable source of information. The value of historical data is distinct from the value of more transient transactions, such as prior authorizations, which, once completed, largely lack useful information. This means that extracting the full value from a fully digital records system will require a significant investment in converting older paper-based records to the required format.

Electronic transactions are not always completely electronic from start to finish. As demonstrated by the CAQH studies, the most frequently used method of communication is mail or fax. After a digital signal has been converted to analog, its usability is significantly impaired. When it comes to document-based data, this entails scanning paper and lossy OCR conversion of images to text.

Portals are a starting point, but not the end. Numerous “digital solutions” actually provide only a partial solution to the problem of creating a digital transaction, preferring to hand off attachment-related issues to human staff. While providers may receive a slick web-based application through which to enter their PA requests, they are essentially replacing the fax machine with a process of extracting medical records, scanning them, and then importing them into the payer system. The payers must then decipher the files and perform laborious data entry.

There is Hope

I am not going to use a tired cliché like “machine learning to the rescue,” despite the significant role it plays. However, there is hope in the form of document automation powered by machine learning. However, even the most sophisticated deep learning systems will fail if the document handling workflows do not change as well. Next time, we’ll discuss how processes can be adapted to fully utilize technology while minimizing disruption.

To learn more, visit

Continue Reading