Friday, June 3, 2011

New Research on EHR and CDS Effectiveness

The following review illustrates some of the methodological difficulties common in current research on EHR and CDS effectiveness.


Health Records and Clinical Decision Support Systems: Impact on National Ambulatory Care Quality
Max J. Romano, BA; Randall S. Stafford, MD, PhD
Arch Intern Med. 2011;171(10):897-903. doi:10.1001/archinternmed.2010.527
This article is available at : http://archinte.ama-assn.org/cgi/content/abstract/171/10/897

Background  Electronic health records (EHRs) are increasingly used by US outpatient physicians. They could improve clinical care via clinical decision support (CDS) and electronic guideline–based reminders and alerts. Using nationally representative data, we tested the hypothesis that a higher quality of care would be associated with EHRs and CDS. 
 Methods  We analyzed physician survey data on 255 402 ambulatory patient visits in nonfederal offices and hospitals from the 2005-2007 National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey. Based on 20 previously developed quality indicators, we assessed the relationship of EHRs and CDS to the provision of guideline-concordant care using multivariable logistic regression.
Results  Electronic health records were used in 30% of an estimated 1.1 billion annual US patient visits. Clinical decision support was present in 57% of these EHR visits (17% of all visits). The use of EHRs and CDS was more likely in the West and in multiphysician settings than in solo practices. In only 1 of 20 indicators was quality greater in EHR visits than in non-EHR visits (diet counseling in high-risk adults, adjusted odds ratio, 1.65; 95% confidence interval, 1.21-2.26). Among the EHR visits, only 1 of 20 quality indicators showed significantly better performance in visits with CDS compared with EHR visits without CDS (lack of routine electrocardiographic ordering in low-risk patients, adjusted odds ratio, 2.88; 95% confidence interval, 1.69-4.90). There were no other significant quality difference.
Conclusions  Our findings indicate no consistent association between EHRs and CDS and better quality. These results raise concerns about the ability of health information technology to fundamentally alter outpatient care quality.


The conclusions reported in this study are consistent with many other recent studies, suggesting that there is no consistent relationship between the use of these tools and effective patient care.  However, it is very important to examine the study design before evaluating the conclusions of this research.  Critical points to be reviewed include the delay in publication with respect to data collection, generally limited use of EHRs and CDSs in US physician practice, the choice of patient visits as the unit for statistical analysis, the procedure for defining the sample, and identification of control variables in the regression model.


The topic addressed in the study is of critical importance, but the rapidly evolving technologies associated with EHR and CDS use require more current assessment. Data collected in the period 2005-2007 provides an interesting historical perspective, but may not yield analyses relevant to the current context.  While EHR adoption in the US remains modest compared to some other industrialized countries (as pointed out by the study authors), state and federal expenditure to promote health information technology adoption is significant, and patterns of adoption have changed on some qualitative dimensions due to emergence of new practice arrangements such as ACOs and implementation of policies such as meaningful use.


The unit of analysis for the study is the patient visit rather than the patient, and performance quality was measured as adherence to guidelines by visit rather than by visits associated with a particular patient.  It was assumed that a higher proportion of guideline adherence by visit should be interpreted as higher quality care provided to eligible patients.  The data analysis does not by itself justify this interpretation.  Drawing any conclusion concerning quality of patient care is further complicated by the lack of consideration of patient clinical profiles - apparently because the authors assert that the quality guidelines should apply to any patient except those presenting potentially confounding comorbidities.  (Patients presenting such comorbidities have conveniently been eliminated from the sample. The authors provide the following example of comorbidity resulting in exclusion of patients - and their visits-  from the sample: asthma in assessing the use of β-blockers in coronary artery disease.)  This reductionist methodological strategy does simplify the statistical analysis, but it also seems to defeat the evaluation of CDS in care of patients who should potentially benefit most from its use.


The authors also mention that they have included emergency visits in the study sample because 
"they are a key source of care and a setting in which EHRs have been more widely adopted", while such visits resulting in hospitalization have been excluded.  The authors do not adequately examine the consequences of these exclusions for the sample size or for interpretation of study results.


The methodology designed for analysis of the data in this study presents several critical weaknesses that may explain the lack of significant results.  In analysis of data from NAMCS and NHAMCS and other similar data sets, future efforts are required to assure patient-centered assessment of care quality - taking into account complex clinical profiles as well as health outcomes over time. Such models would be especially useful for longitudinal analysis as new data become available.