In this study of almost 30 million visits within an integrated delivery system, there were dramatic increases in the percentage of office visits with diagnoses available within an electronic database on the same day as the visit during 2004–2006, corresponding to the implementation of increasingly sophisticated forms of HIT for office visit documentation.
Within any health system, there are multiple potential delays in visit documentation: 1) clinician delay, such as when clinical care takes precedence over documentation; 2) clinical information delay, such as when the diagnosis requires clinical information that is not immediately available, e.g., a laboratory result; and 3) systemic delays, such as prolonged processing time after the clinician finishes the record. The increase in timely availability of clinical data with HIT use likely results from decreased clinician delay and decreased systemic delays, since HIT allows for more swift entry of diagnoses or eliminates steps in the data capture process. One important advantage of an EMR over an entirely paper system is the decreased clinical information delay via swift access to laboratory and radiology results, for either the ordering clinician or another provider. This did not contribute in our study since results viewing was available throughout the study period, but may be a contributor in other systems that do not already have electronic access to laboratory values and radiology reports, particularly for care shifted between providers, when paper lab results may not be readily exchanged. Health care systems without such electronic transfer of laboratory, radiology, and consultation data might experience even more dramatic improvements in clinical data timeliness.
The problem of clinician delay may be exacerbated by the presence of sicker patients, which are more time intensive to care for and leave less time for documentation. We did not see any evidence for increase or decrease of acuity in the patient characteristics over time (age and number of diagnoses stayed had minimal variation), so it would be unlikely that a decrease in patient acuity would have contributed to the decrease in time to documentation. If the population had grown sicker during this time, then the effects of Advanced may be even more pronounced than they appear in our analysis.
The differential pattern of timely availability of diagnoses in individual medical centers reflecting the use of Advanced HIT at those centers (Figures and ) suggests that Advanced HIT may drive the timely entry of diagnoses more than Basic or Intermediate HIT. In most medical centers, once Advanced HIT was implemented for 80–90% of patient visits in the office setting, 99–100% of visits had documentation complete by the same day. While it is possible that Advanced HIT implementation increased medical center focus on timely documentation, the gains in timely data entry were coincident in each medical center with the uptake of Advanced HIT, and it is unlikely that every medical center had the same emphasis on timely documentation during the period of Advanced HIT implementation. The relationship between Advanced HIT and the more timely availability of information compared to Basic HIT or Intermediate HIT implies that different HIT functionality may lead to different effects on timely availability of clinical information, and that the systems that allow the most streamlined workflows may lead to the most timely information availability.
In other health care systems, individual providers may be motivated to complete diagnostic coding in order to get reimbursed in a more timely fashion. The results of our study, if applied to other medical providers using an EMR, suggest the possibility of timelier billing cycles. In contrast, KPNC is a prepaid, integrated delivery system, and while recording diagnostic codes is considered a part of standard clinical documentation for quality and organizational reasons, compensation is not dependent on entry of diagnostic codes. Therefore, improved billing cycles would be an unlikely driver in the increased timely availability of data demonstrated in this study.
In this study we did not find a change over time in the number of diagnoses for visits (Table ). It has been hypothesized that an EMR would increase the number of diagnoses attributed to any patient, leading to improved data "completeness." A possible reason for improved completeness would include greater ease of documentation with an EMR.[32
] It may be that the number of diagnostic codes entered did not change because the ones that were documented on paper records represented the complete clinical picture. It may be that there was no motivating factor for providers to increase the numbers of diagnoses charted in the EMR. We do not have information regarding this clinician behavior but it may be amenable to collection through qualitative or quantitative physician surveys.
It is possible that some degree of change in the timely availability of diagnostic data results from changes in patient population or practitioner population over time rather than changes in HIT use over time. Though we did not adjust for possible confounders such as practitioner characteristics, patient characteristics, or seasonal variation, Table demonstrates the stability of patient and practitioner characteristics over time. This provides evidence that any contribution they make to the change in availability of diagnostic data is likely to be small and not sufficient to explain the substantial increase in percentage of same-day availability of diagnostic data.
This study does not address the accuracy of the diagnoses entered. Errors in diagnostic accuracy may occur in two forms: diagnoses incorrectly attributed to the patient (e.g. assignation of hypertension when the patient has never been hypertensive); and diagnoses incorrectly attributed to the patient at the time of charting (e.g. anemia is recorded as a diagnosis even after it has resolved). Advanced HIT may facilitate the second type of error since it enables carryover of information from one visit to the next without a forcing mechanism to verify the information. The use of SNOMED codes, which allow for much more detailed specification of diagnoses than ICD-9 codes, may enable a more structured description of a visit, and therefore a more accurate one; demonstrating this was beyond the scope of this study but could be considered in the future.
The study occurred in a single, integrated delivery system. While the number of visits is quite large, findings may be specific to this patient population, practitioner population, or particular system, and may change with different HIT systems, or in non-integrated health care delivery systems. However, the Advanced HIT EpicCare© system is commercially available, and therefore our findings provide information that may be useful to others considering implementing it in their own settings. Importantly, there was substantial time allotted to implementation of the Advanced HIT system to allow for adequate initial and ongoing clinician training, including policies such as decreased patient load for clinicians during the initial implementation at a site. This careful implementation and ongoing support may be more difficult to achieve in a smaller or non-integrated setting.
It was beyond the scope of this study to examine associations between changes in clinical outcomes and the improved timely availability of clinical information. However, documenting the effect the time to availability of clinical information has on the quality of care delivery and on individual outcomes would aid in quantifying the contributions EMRs make to improvements in overall health and we hope to examine these relationships in the future.
Finally, the study health system has collected electronic clinical data since well prior to the study period, so results may be different for organizations that have not previously collected electronic data. The change in timely availability with use of a HIT system may be more extreme for organizations that are still entirely paper-based.
The current public health infrastructure is limited and suffers from a dearth of clinical information. Few public health officials receive routine, timely transfers of electronic data, have mechanisms to analyse the clinical data, monitor outbreaks, or even confirm other reports. Most public health localities rely instead on word of mouth reports, phone calls with clinicians, or in-person interviews.[8
] Compounding this problem are the limited and often declining resources available for public and population health.
Electronic data captured by HIT systems offer tremendous promise in improving the availability of timely clinical information for disease surveillance, responses to potential outbreaks, and monitoring of actual outbreaks. For example, an electronic surveillance system implemented in laboratories in two Indiana counties led to a 29% increase in absolute number of Shigella
species infections identified during an outbreak, and led to same day notification to the health department, decreased from an average lag time of 2.5 days from time of positive result to time of health department notification.[34
] We have demonstrated that in an integrated delivery system using a comprehensive EMR, we can capture multiple different types of clinical diagnostic information, not just infectious disease entities.
In addition, HIT systems may improve quality of care by providing timely clinical information at the point of care. Patients with chronic disease require coordination of care across multiple practitioners, including primary care providers, specialists, care managers, pharmacists, and others. Relevant patient information often is not available at the time of the visit, which impedes clinical care.[20
] A large survey of primary care physicians in Colorado, most using paper charts and a few using full electronic records, found physicians reported missing critical information in 13.6% of visits. The physicians also reported that the missing information resulted in potential delayed care or additional services for 59.5% of those visits.[20
Patients who would have most likely benefited from the improved information availability in the KPNC system are those who re-present within one week of their visit. It has been shown in other settings that 6.4% of asthma patients seen for urgent care follow-up within one week.[35
] We found previously that in 2002, KPNC members made 2.1 million visits for asthma (unpublished data), which implies up to ~133,000 repeat visits of an asthmatic within one week of initial presentation. Other common chronic diseases such as congestive heart failure, hypertension, and diabetes also may lead to repeat visits within a short time frame. It is these patients who will likely most benefit from HIT improvements in timely availability of information through processes such as medication management, coordination of care, and communication between providers.