Sunday, November 18, 2007

HIT before HIE (and not without my knowing what’s out there about ME)

There is a grand vision out there in the world of health information technology. It goes something like this.

One day, sometime in the future, every time I encounter the health care system, the details of that encounter will be recorded on a computer. This includes everything a health care provider writes (er… types) about me in their electronic health record system (EHR), every laboratory, x-ray or other diagnostic test result, every prescription ever given to me and information from the pharmacy on whether I got it filled and refilled. All of these computers will be connected through RHIOs (regional health information organizations) and all the RHIOs will be connected in some manner to the National Health Information Network (NHIN). The end result will be that anywhere in the country that I seek health care, with my permission, my treating provider will be able to search this network, find all the information stored about me, have this information consolidated and organized into a useful format and then use this information to help plan my treatment. The promise is that this will have dramatic effects on the quality of care I receive and the overall cost of medical care in this country will decrease.

This vision is driving an enormous national effort to set standards for how these computers will “talk” to one another, to establish rules for the privacy and security of the information and to figure out the path to bring this dream into reality.

The vision is grand and I am sure will be realized some day. But there is a problem. There is a problem in focusing our efforts on connectivity, interoperability and exchange of information when the vast majority of that information is not electronic today. There is a problem in investing in a technology and an idea on a national scale without a substantial portfolio of regional and statewide efforts to build upon. But most importantly, we are overlooking many, less expensive and better proven methods of improving the quality and safety of medical care in the United States. The dream of a National Health Information Network (NHIN) is worthy of attention but it is a dream we should put off for another night.

Health information technology in the hands of health care providers has great potential to improve quality, improve safety and improve communication with patients. With limited funds available for technology development, we should be spending the vast majority of our IT resources now in supporting the implementation of health information technology (HIT) rather than the future vision of widespread health information exchange (HIE).

It is estimated that the increased hospital costs alone of preventable adverse drug events affecting inpatients are about $2 billion for the nation as a whole. The table below is referenced in the IOM report To Err is Human and seeks to create a taxonomy of medical errors. This is useful in examining which, if any of these errors would be alleviated by health information exchange. Conclusion: very few of them would be affected beneficially. Which of them would be improved by health information technology implemented in the provider sites? Almost all of them.

HIT before HIE

Here are just some of the reasons why we should put our resources into HIT, before HIE

  • Electronic access to an individual’s health information from another provider’s EHR system will not address the major ways in which they are likely to be injured by the health care system
  • Almost all documented quality improvements and safety enhancements through health information technology have come from the use of HIT systems within an organization
  • There are no operational studies which demonstrate a reduction in health care costs through health information exchange – the often quoted models are based on dozens of unproven assumptions
  • With the low prevalence of electronic health records and no mandate for providers to make this transition, many providers will continue to have records that are not plugged into a data exchange. The result is that not everyone who wants or needs their record in the NHIN will have it there if their provider is not connected.
  • With incomplete information, much of the information contained in the NHIN (or another data exchange model) will be incomplete and will not reflect the patient’s current status
  • There is no way for a treating provider to validate the quality of the information provided by other treating entities. Without validation it is possible that inaccurate information is passed from one source to another without critical examination of its validity.
  • Duplication of some studies – x-rays over time, serial lab tests and others – may in fact be optimal medical care. Assumptions have been made that this duplication is expensive and will be largely eliminated when all information is available on the NHIN.
  • Doctors who make money by doing tests may perform duplicate tests even if prior information is available, questioning the validity, timeliness or quality of prior testing.
  • Pharmacy data retrieved will tell providers what was prescribed, maybe even tell what was filled at a pharmacy, but cannot tell what medications the patient had stopped by another provider or whether the patient is actually taking the medication. It also won’t tell if the patient has started using old medications around their house or meds from a family member or friend. Nothing substitutes for reviewing a patient’s medications and having them bring the ones they are taking for review by the treating provider.
  • Current technology cannot provide absolute privacy and security and people need to be convinced that when their privacy and security risks are balanced against the potential benefits of having their health information on the NHIN, that the latter is seen to be dominant. There is survey data to indicate that many patients will voluntarily agree to release their information to their local data exchange and thus, become part of the NHIN. There is reasonable evidence however that patients will not want to have any information about them shared without having access to that information themselves. (“Nothing about me, without me”)
  • There are no studies that have looked at the errors that might arise during the decades that it may take for this ultimate vision to be realized – during which time information on patients opting in to the data exchange will be largely incomplete, and thus inaccurate
  • While some estimates have been made of the costs that will be added to each clinical encounter through maintaining, querying and contributing to the NHIN there are no proven business models for sustaining these costs going forward. Health care costs in this country are already higher than anywhere else in the world and we have not yet dealt with the costs of health information technology and health information exchange.

Getting to widespread adoption of HIT and then HIE

Problems notwithstanding, the ultimate vision of the NHIN is the right one. The path we have set to achieve it is wrong. We are funding HIE projects for implementation when there are not standards in place for how privacy and security will be managed. Current certification (by CCHIT) of EHRs does not require that interoperability standards have been met. We are funding a weed garden of RHIOs, each exchanging different sets of information, each with different participation agreements, different platforms, different methods of patient identification and not all using the same interoperability standards. Some providers, because of either functional or geographic overlap, are involved in two, three or more RHIOs. While this may make sense today, ultimately it is wasted effort and will surely not make sense in the future. I think the logical rollout of HIT and HIE should look more like what follows:

First Order Priorities

  • The vast majority of development resources at this time should be directed to health information technology (HIT) implementation in all locations and a timeline should be set for providers to be required to have installed a fully functional EHR that meets the certification requirements of CCHIT (the Certification Commission for Health Information Technology)
  • Priority funding should be given to demonstrations of improvements in quality and safety using HIT within hospital and ambulatory care systems with rapid dissemination of models that work
  • A certification program like the one being developed by the National Committee for Quality Assurance (NCQA) should be implemented immediately to set the goals for HIT adoption and use. This certification should be used to establish a fee enhancement that all insurers (public and private) would pay and which could be used by providers to support improved quality of care

Second Order Priorities

  • CCHIT criteria should include requirements for a patient portal so that patients are able to have access to all of their health care information. This is an important precursor for the implementation of full information exchange. People need to know what information is in their electronic health files, have the opportunity to validate this information and feel secure that the information is secure
  • Priority should be given for connectivity of EHRs to a public health syndromic surveillance system. With fears about an influenza pandemic raising year by year, such connectivity could provide life-saving early warning enabling an increased lead time for dispensing immunizations and anti-viral medications.
  • Standards for interoperability must be developed and vendors should all be required to adopt these to become CCHIT certified

Third Order Priorities

  • Once a national framework has been developed for privacy, security and interoperability standards and appropriate legislation is in place the NHIN should be implemented through the vendor community. Each vendor would serve as a node in the NHIN, implementing the privacy, security and interoperability standards and linking their users into the national framework. This would completely alleviate the RHIO madness that we are in and the costs of connectivity would be borne by the provider community to be reimbursed through national requirements for private insurers to pay for connectivity. Hopefully, when HIT is completely implemented and the NHIN is a reality, the overall cost savings in reduced hospitalization, reduced medical errors and increased patient safety should offset the costs of the technology and may stabilize or even reduce health care costs overall.
This model makes logical sense to me. It has significant advantages over the obsession we seem to be witnessing with adopting interoperability today. We need to refocus our efforts and our funding on widespread adoption of health information technology today and leave the promise of a web of connectivity until tomorrow.

Sunday, November 4, 2007

Community Health Centers in the Service of Public Health

For too many years public health departments and community health centers have lived in their own worlds, with public health focused on geographically and politically designated populations while community health focused on the delivery of primary health care services.

In 2002, when the Institute for Family Health went live with the EPIC electronic health record and practice management system one of our first major projects was to link with the New York City Department of Health and Mental Hygiene to eliminate these silos and to explore the many ways that our work was complementary. What has resulted is a significant array of joint projects that was honored with this year with receipt of the 2007 Davies Award in the category of Public Health from HIMSS. A few of the figures from our application are posted below. We are very committed to using the electronic health record to link our community health center network to the important work of our health department in monitoring and improving the care of all New Yorkers. (You can double-click on any of the figures to enlarge them for easier reading).

Figure 1
The New York City Department of Health and Mental Hygiene collects data from a variety of sources for its Syndromic Surveillance System including ER data, 911 data, absentee data, and pharmacy data. In this figure, the results of their ER data collection are compared to data they receive from our EHR using fever and respiratory symptoms as the trigger. It can be seen that the Institute's EHR data peaks days before the ER data when Flu A is prevalent and is much more responsive than the ER data in showing the less serious flu B outbreak that appears later in the year.

Figure 2
This figure represents the bilateral transfer of information between the NYCDOHMH and the Institute for Family Health. When the NYCDOHMH receives a signal from any source of an outbreak they immediately issue a bulletin via email to all NYC providers. That notice is picked up by our Chief Medical Information Officer and if appropriate an alert is programmed immediately into our EHR. This alert may ask that Insititute providers collect specimens for the DOH for further investigation.

Figure 3
In September of 2007 the DOH reported an outbreak of Legionairres' Disease in the Parkchester neighborhood in the Bronx. Alerts were immediately put in the system such that when a patient presents in either of the two centers we operate in Parkchester, and has a chief complaint of cough the provider receives a message to consider Legionella as a possible diagnosis.

Figure 4
The alert is coupled with a Smart Set of possible notes and orders to facilitate the proper response by the provider

Figure 5
In another shared activity with NYCDOHMH the Institute has begun tracking outcomes on over 40 indicators which are related to the City's "Take Care New York" program. In the example below we track the percent of patients (Men > 35, Women > 45) who have met criteria for cholesterol testing.

Figure 6
In this figure we are tracking the progress that results from an alert to nurses and providers to update the substance abuse history portion of the EHR. One can see that compliance was quite poor to begin with but has steadily increased at all sites.

Additional joint projects include
  • producing a series of algorithms for decision supports that are generic and will be in the public domain and are specific to primary care practice
  • a bilateral interface with the immunization registry
  • automated reporting of communicable diseases
  • development of a "model" EHR for public health

Thursday, November 1, 2007

HIT and HIE for America’s Most Vulnerable Patients

Today I will testify before the U.S. Congress, House of Representatives, Committee on Oversight and Government Reform, Subcommittee on Government Management, Organization and Procurement. The hearing, entitled “Too Many Cooks? Coordinating Federal and State Health IT” could go on for weeks if the issue were to be thoroughly explored. I was asked by the Subcommittee Chair, Edolphus Towns to present our experience with health information technology in community health centers and make some recommendations.

The rollout of HIT/HIE must include special considerations for vulnerable, diverse populations or run the risk that it will increase disparities in health services delivery and health outcomes. I will make the following recommendations at the hearing.

Insure that all legislation that supports the implementation of electronic health records targets those patients at highest risk on our society. This requires that all developments have organized ways for community participation in their planning and execution and that providers in the health care safety net – community health centers and public hospitals – are included as primary targets for funding implementation of health information technology

Insure that all legislation that supports health information exchange (HIE) includes provisions that safety-net providers – specifically community health centers and public hospitals – must be included in HIE networks. This requires that they first be given the resources to implement electronic health records (because you need an electronic health record in order to participate in a health information exchange). This also requires specific funding for implementation and maintenance of connectivity to HIE networks for safety net providers who may otherwise be unable to pay for the needed interfaces of their systems with these networks

Fund the integration of all EHRs, but specifically those in safety net facilities, into local, regional and state health department systems that monitor the disease burden in the community and help people optimize their preventive health care measures. Medically vulnerable populations are often the first victims of infectious diseases as they also suffer from inadequate nutrition, housing, clothing, health education and access to medical care.

Mandate that all EHR systems capture data on race, ethnicity, gender and primary language so that providers can examine disparities that exist in treatment and in outcomes within their health care delivery systems and work to correct them through targeted efforts aimed at the highest risk populations

Encourage EHR vendors to provide a mechanism for alerting providers to clinical trials which may be relevant to their patients. People of color are underrepresented in many clinical trials in this country and EHRs provide an easy way to identify patients who meet clinical trial criteria so that they may be offered the opportunity to participate. This improves the relevance of clinical findings of these studies to ethnically diverse populations.

Create criteria for EHR certification that require that information produced for patient consumption such as health education materials, visit summaries, portals that allow patients internet access to their own health information – all be made available in multiple languages and at a 4th grade literacy level when needed.

Insure that rural areas are funded to deploy broadband technology and that broadband access is provided in all public housing being built. A timetable should be set to retrofit all existing public housing facilities with broadband capability.

Require the input of communities of color in planning the privacy and security requirements for health information exchange and in developing consent procedures for participation in health information exchange. There are well documented, legitimate reasons that issues of privacy, security and consent procedures will play out differently in communities of color. Among other issues, it must be made clear that information provided by patients and entered in their electronic health records as well as information shared in health information exchange networks is not subject to government use for purposes of identifying undocumented immigrants.

Provide resources for health information technology adoption in prisons, in the foster care system, for homeless health care providers, for migrant health care providers and for other providers serving vulnerable populations so that these mobile and transient patients can have records that are available wherever they may go.

Create a national system for specifically monitoring the impact of health information technology and health information exchange efforts on vulnerable populations and tying future funding to successes in eliminating disparities in treatment by the participating health care providers and disparities in the outcomes of their patients.

Insure that any pay-for-performance systems include mechanisms for risk adjustment for outcome evaluations – not to reduce the goals that we must reach but to recognize the added difficulty that traditionally medically underserved people may have in achieving these goals.


I look forward to suggestions and enhancements to the items cited above.