ConductScience Digital Health

CDS Hooks

About CDS Hooks

As a clinician is using the system, the EHR may fire off a notification. The notification is normally sent to an external decision support service. When writing a prescription, the service learns that the clinician is in the process of prescribing a drug, and it can return some information such as a new proposal usually displayed as a card inside the EHR. For instance, if the clinician is prescribing the brand-name drug, the information returned by a CDS service is relevant to the task at hand. The card that is displayed can propose a different drug that is more effective, cheaper or available at the moment. If the clinician likes the card returned by the CDS service, he can just accept it by clicking a button.

The service can also provide a link to an external app. At some point in the workflow, if the clinician may be required to run an antibiotic selection application, the CDS Hooks can return a link to that app. CDS Hooks is an important complementary technology to Smart apps because at times the clinician might forget about it altogether, or might not even be aware of the external app, but the system reminds them about it as the card pops up waiting for their approval to execute the process. CDS Hooks helps the clinicians run the right app that is required at a given stage, and this can also save many explicit steps.

There are three types of cards that are embedded in the EHR, each displaying different information:

Information card: This conveys text that may be useful for the user.

Suggestion card: This card provides alternative suggestions.

App link card: This card links to other SMART apps or reference materials.

Documentation

Introduction

SMART on FHIR has been a game changer in health care by lowering the barrier to innovation which has led to the development of an ecosystem of apps to meet various needs. With SMART on FHIR, structured healthcare data is accessible by apps in a variety of EHR systems; as a result, different apps have been integrated into several EHR systems.  A good number of EHR vendors have already implemented support for the protocol as various third-party applications have already been plugged in. The issue that informatics experts familiar with FHIR are now dealing with is how an end-user can know which app to use and when. CDS Hooks seeks to empower clinicians to know what apps they should be running at a given time in their workflow. For instance, if a person wants to run an app that can help in changing the dose of a patient based on his genotype, it would be necessary to invoke that app while making the prescription. CDS Hooks runs these automated checks ahead of time for clinicians; presenting the relevant advice only if and when it matters within the context of the EHR.

Challenges of the CDS system

Like any other system ever developed by man, certain aspects have been abhorred by the users. Many physicians have complained that the CDS system has too many alerts, many of which may not be relevant. Having many irrelevant and intrusive alerts within EHR systems can greatly interrupt the workflow.

CDS Hooks is still less developed as compared to the SMART on FHIR protocol. Unlike the SMART on FHIR which has been around for some years now, the CDS service is still at a much earlier stage. The SMART on FHIR protocol is much more developed regarding the security of the data to the vocabulary used. With CDS, there are still many aspects that still need some more engineering work; how to launch the apps and the security model have not yet been thoroughly defined. The basic workflow has however been streamlined; how a system asks for advice and how that advice is presented have already been defined.

The trick here is in letting the real requirements from EHRs and CDS vendors be the basis that drives work in this sector instead of coming up with custom solutions to certain challenges. Custom-made solutions can work in certain contexts, but they may not be applicable in all cases. At the moment companies that build decision services, EHR vendors Epic and Cerner are actively involved in the project.

The solution to these challenges

  • Cards should be non-modal and be designed in such a way that they do not affect the workflow or get in the way.
  • EHR can also have policies such that physicians have a say in the matter. In this case, when a user sees a card that does not apply or that they do not care about, the user can select the option- do not show me cards like this anymore. In this case, the EHRs can take into account such feedback from the users when displaying the cards.    

CDS Hooks has the potential to increase the number of decision-support vendors. A deep ecosystem would result from this as different apps that provide different services would emanate with CDS seeking to integrate or aggregate their results. Clinicians would then be empowered to choose the services that suit their practice best. There is also room for aggregation services to come in and manage some of the logic of analyzing the results for the user.

Hospitals would have to decide whether or not a given service is appropriate and worth buying or using. Some things are however much easier to test on CDS Hooks than others. If for instance, a hospital wishes to prescribe a service that can detect and flag drug-to-drug interactions, there’s a simple and straightforward way to test the CDS Hooks interface. In such a case there happens to be a simple API where these hooks can be applied. To write tests that can help define the efficiency of such a system, all you need to do is to test different drugs that you know should interact. You can carry out such tests routinely employing different drug types to get assurances that there has been no change in the responses from the CDS service, as it has become evident that there can be a change in the behavior of the system especially when the system is upgraded or modified.

CDS services add much value to health care by lowering the cost of the integration and as such developers can focus more on creating apps that add value to the systems. Reducing the barriers to technology integration means that developers can focus more on designing great services and creating efficient market strategies.  It is possible to envision a CDS services marketplace where such services would be free or shareable. To the extent that the kinds of advice and tools built up by some academic centers and hospitals are through government funding or considered a public good, an ecosystem where such tools can be shared widely is feasible. It is, however, a known fact that academic medical centers often build technology in-house and then later try to commercialize it.

Case Study

About 17.9 million U.S. residents were diagnosed with diabetes in 2007. The total cost for this diabetes-related medical spending was about $174 billion. The Centers for Disease Control (CDC) has estimated the lifetime risk of developing diabetes for individuals born in the United States in 2000 to be 32.8 percent for men and 38.5 percent for women (Narayan et al. 2003). By 2034, the number of U.S. residents with diagnosed or undiagnosed diabetes is projected to increase to 44.1 million, accompanied by $336 billion in annual diabetes-related medical spending (Huang et al. 2009). New strategies are required, both for diabetes screening and prevention (Gilmer and O’Connor 2010). Billions of dollars have been invested by various medical groups into electronic medical records, but there have not been sufficient studies that have determined the cost-effectiveness of EMR-based clinical decision support. Integrated CDS systems have the potential to improve clinical care as well as reduce the costs associated with treating the millions of patients that are enrolled in health plans that have deployed EMRs.

EMRs can be programmed to include sophisticated algorithms that exploit current and past clinical information to provide detailed recommendations at the time of a clinical encounter (Von Korff et al. 1997). One of the biggest challenges in diabetes care in patients that have not yet attained the recommended clinical goals is the lack of timely intensification of pharmacotherapy. Medication non-adherence and other competing demands at the time of the visit may be some of the reasons for this. Some of the attempts that have been made to improve this include telephone-based management, team-based case management as well as information technology-based interventions.

EMR-based CDS treatment has been tried, though unsuccessfully, in the treatment of diabetes and some other chronic conditions such as congestive heart failure (CHF), hypertension, and asthma.

Some of the reasons why the use of EMR-based CDS cares initially failed include:

  • CDS displays were availed late and were often not viewed or skipped by the primary care physician as opposed to being employed during visit planning.
  • Most of the CDS did not include detailed drug-specific advice but was reduced to general prompts and reminders.
  • The introduction of CDS service was not fully maximized as it was not accompanied by other changes in the clinic workflow and other staff responsibilities.

To address these concerns, an improved model of EMR-based CDS was employed. It addressed these concerns by:

  • Reorganizing the workflow so as to include guidance in visits-planning activities.
  • Giving treatment recommendations that included detailed and personalized drug-specific advice.

A study of the cost-effectiveness of the EMR-based CDS system in the treatment of diabetes was carried out in the Midwestern health plan, and it revealed that EMR-based CDS health care is not only cost-effective in the acceptable standards, but it was also valuable in the sensitivity analyses. This summed up the importance it can have in health care. The observed clinical impact is comparable to that achieved by many disease management or patient education programs that are more expensive (Norris et al. 2002).  As a matter of fact, a major appeal of the use of EMR-based CDS interventions in clinical care is its potential to be deployed over a large population at a very affordable cost.

The adoption of these tools will, however, require incentives for provider participation and some changes in provider behavior. In the Midwestern health research study, the intervention was carried out on a research basis, but if the approach is established and adopted widely, such incentives to providers may become less important.

Conclusion

CDS-Hooks on SMART on FHIR facilitates SMART CDS apps to fire off some specific notification card to the supporting EMRs. CDS-Hooks, such as SMART, is an open standard that is tailored to integrate seamlessly with the diverse EHR vendor systems. The service will register for the hooks that they wish to respond to and send CDS cards to trigger the necessary actions. The cards present tailored patient-specific information, simplifying the job of the end user (clinician). EHRs allow the interested CDS services to register on the diverse CDS hooks, directing the various actions as necessary. EMR-based CDS health care is scalable and can be adopted alongside other additional and complementary care improvement strategies. More research is necessary to deepen our knowledge of the clinical decisions that can be supported by EMR-based CDS and the instances in which such would be cost-effective.

In the brave new world of personalized medicine and mixed clinical outcomes, strategies that can personalize and standardize clinical care are becoming increasingly important. Tailoring the treatment goals for individual patients is becoming increasingly important. For instance, a particular treatment strategy may be increasing the risk of mortality while attempting to reduce microvascular complications among diabetes patients. Clinical decision-making and the costs associated with this are becoming more challenging and complex. CDS strategies are proving important in this regard, and more investments are required to improve the effectiveness of this technology.

References

  1. Narayan, K. M., J. P. Boyle, T. J. Thompson, S.W. Sorensen, and D. F.Williamson. (2003). Lifetime Risk for Diabetes Mellitus in the United States. JAMA 290 (14): 1884–90.
  2. Huang, E. S., A. Basu, M. O’Grady, and J. C. Capretta. (2009). Projecting the Future Diabetes Population Size and Related Costs for the U.S. Diabetes Care 32 (12):2225–9. doi: 10.2337/dc09-0459.
  3. Gilmer, T. P., and P. J.O’Connor. (2010). The Growing Importance of Diabetes Screening. Diabetes Care 33 (7): 1695–7. doi:  10.2337/dc10-0855
  4. VonKorff,M., J.Gruman, J. Schaefer, S. J.Curry, E. H.Wagner. (1997). Collaborative Management of Chronic Illness. Annals of Internal Medicine 127 (12): 1097–102.
  5. Norris, S. L., J. Lau, S. J. Smith, C. H. Schmid, and M. M. Engelgau. (2002b). Self-Management Education for Adults with Type 2 Diabetes: A Meta-Analysis of the Effect on Glycemic Control. Diabetes Care 25 (7): 1159–71.

Learn more about

ConductScience Digital Health

Let's work together!

Have questions? Ask anything!