The problem list is a key and required [1] element of both the paper record and EMR. Weed [2], [3] popularized the use of the problem list in an influential book and papers. Benson et al [4] have argued that the “problem list and the medication record are particularly useful for providing an overview of patients’ significant diagnoses and treatments. If well-structured, reliable, and consistent, they can also contribute substantially to the quality of patient care.” An accurate problem list facilitates automated decision support, clinical research, data mining, and patient care [4], [5], [6]. The Joint Commission on the Accreditation of Healthcare Organizations [1] mandates maintenance of a problem list. The problem list can be a useful tool in both paper records and EMRs for organizing physician notes, making patient rounds, and “signing out” patients to covering physicians [7], [8].
The advent of EMRs could significantly increase the potential power and importance of the problem list. Database links and pointers could make the problem list a true “index” to the EMR as originally conceived by Weed [2], [3]. For example, selecting a problem off the problem list (e.g. hypertension) could link to all relevant healthcare encounters for hypertension, all relevant laboratory results, and radiology reports that mention the diagnosis of hypertension, or link to all drugs that treat hypertension. Furthermore, detailed electronic problem lists can facilitate the development of CDS, patient registries and research. There is little information on the accuracy of problem list maintenance at either teaching or non-teaching hospitals. A limited study in the adoption of EMRs revealed that users produce more complete problem lists in an EMR than in the paper medical record [9]. However, in a study of the Veterans Administration EMR [10], the sensitivity of the electronic problem list for the diagnosis of hypertension was only 49%. Campbell [11] has attributed poor clinician compliance with the problem list to “minimal rewards for the clinician”.
In an audit of 105 outpatients seen in the Neurology Clinic at the University of Illinois Medical Center in 2002, 19% of the medical records had no problems of any kind listed on the electronic problem list and 39% had no problem on the electronic problem list related to the current visit [12]. Furthermore, 47% of the problems were entered as “free text” rather than as codified discrete ICD-9-CM [13], ICD-10 [14], or SNOMED® [15] codes. In an audit of inpatient medical records we compared the problem list as reported in the medical record with the problem list as ascertained by chart review [12]. The number of problems on the problem lists was roughly only a quarter of those found by auditing the chart, while 46% of the charts had an empty problem list. Problem list under documentation is certainly not unique to our institution, a recent study from intermountain health care also reported that their problems lists are “Usually incomplete and inaccurate, and are often totally unused” [16].
One of the potential causes of poor problem list documentation is the use of controlled terminologies; ICD-9-CM [13], ICD-10 [14], or SNOMED® [15], which may not be ideal for documentation of problems in a problem list [17], [18]. There are many reasons that problem list use may be underutilized [17], [18] which may help explain the interest in “free text” problems by users. One of the strategies to improve the use of standardized terminologies has been the development of information technology to convert non-standardized problems into local controlled terminologies [16], [19], [20], [21] as well as standard terminologies such as SNOMED® [15], [22].
More accurate problem lists may lead to higher levels of patient safety and lower levels of medical error. Carpenter and Gorman [23] have tested a natural language processor that detects medication errors based upon a mismatch between the drug ordered and problems that reside on the problem list. For example, if a drug does not match up with a problem, either the problem list is deficient and needs to be updated or the drug has been ordered in error and needs to be deleted. A theoretical CDS prototype [24], [25] has been tested and suggested that integration of problem list maintenance into CPOE workflow may promote better problem list documentation. In this study, we have tested a CDS system that helps maintain the electronic problem list in a real-time clinical environment. The CDS system is triggered by drug orders in CPOE to generate alerts to providers who then have the option to update the electronic problem list by a simple automated process.
The University of Illinois Hospital, a 450-bed teaching hospital, utilizes a commercially available EMR (Millennium®, Cerner Corporation, Kansas City, MO) which is used as the primary repository for all results, problem lists, clinical notes, medication lists, and orders. All inpatient medication orders are placed using CPOE. The commercially available CDS (Discern Expert®, Cerner Corporation) has been previously described [26], [27], [28].
Our problem list is multidisciplinary, allowing any
The total number of alerts for all target diagnosis groups in their 2-month sample observation periods was 1011 (Table 2). Of the six categories of alerts, the hyperlipidemia-coronary atherosclerosis alerts fired most often followed by the diabetes alerts. The overall alert validity was 96 ± 1% (Table 2). The highest levels of validity for the alerts were for diabetes and hyperlipidemia, the lowest levels of validity were for ischemic stroke and HIV (Table 2).
When triggered, alerts led to an
We have demonstrated the feasibility of using CDS driven by medication orders from CPOE to assist in the maintenance of the problem list. Problems can be added to the problem list during the routine work of placing medication orders with less clicks than would otherwise be needed in the EMR studied.
We believe that this CDS system has been well received at our institution due to the relatively high compliance rate. This is likely due to the reduction in the labor, or number of screens and clicks
This analysis was limited to inpatient order entry with CPOE. Although we use the same EMR in the ambulatory environment, electronic prescribing was not fully implemented in the ambulatory setting to test whether outpatient prescribing would be similarly successful in generating additions to the problem list. In addition, our institution is a public teaching hospital with the majority of orders placed by housestaff physicians using CPOE. While the CDS could be implemented regardless of the type
In summary, this novel CDS worked within the process of medication ordering at our institution. It was successful in improving problem list documentation with minimal diagnostic inaccuracies. Future work should be focused on expanding this work to more medications as well as for use in all forms of clinician orders to increase the scope of problems that can be added to the problem list through this type of CDS coupled to CPOE.
The authors have no financial conflicts to disclose. Summary points What was already known in this field? Problem lists can be useful in EMRs as they have been used to help automate note entry, sign out reports, rounding reports and clinical decision support. Compliance with problem list documentation is a problem. The utility of problem lists in EMRs is limited due to the non-compliance with documentation. What this study has added to our knowledge: This was a novel type of CDS based on voluntary,
With special thanks to Lisa Canonge, Marla Lax, R.N. Amy Looi, R.N. and Jennifer Welch C.T. (ASCP) for assistance in the CDS alert development.
This study was approved by the University of Illinois at Chicago Institutional Review Board.
The study was in part funded by the National Patient Safety Foundation, James S. Todd Memorial Research Grant (PI: C. Jao). The funding source had no role in the production of this manuscript.
Dr. Galanter was supported by grant number U18HS016973 from the Agency
A total of 91% of subjects with AERD reported that aspirin therapy was beneficial but less than 50% of the survey population had been offered aspirin therapy.6 The present algorithmic approach could be used to display automatic alert notifications to physicians to promote the consideration of AERD and improve documentation of AERD,26 while offering evidence-based information and detailed advice including referral options.27 Providing patients with an accurate diagnosis may empower them to seek out effective treatments for their disease and/or engage in clinical trials that have the potential to transform the future of AERD-specific care.
This would provide emergency physicians with relevant information and help them keep the medical record up-to-date, with minimal effect on existing workflows, as medication reconciliation is generally performed for all emergency department visits, particularly those who are admitted. Previous studies have shown that discrepancies between medication lists and problem lists can help identify patients who have been treated for type II diabetes but have not been formally diagnosed in the medical record [6], and that automatic reminders to providers based on the medications they order can help to improve their documentation of the problem list [7]. To do so, we selected a group of sentinel conditions – obstructive lung disease (asthma and COPD), hypertension, diabetes, thromboembolic (TE) risk (prior venous thromboembolism or high risk condition for embolism), and congestive heart failure (CHF) – which are important to both emergency medicine and primary care, and which are at particular risk of iatrogenic complications within the hospital setting if they are not recognized, such as a patient with unidentified CHF receiving a rapid fluid bolus, causing pulmonary edema and respiratory failure.
