Quantifying Imaging’s Value to Patients
Contrary to popular belief among regulators and payors, imaging studies do, in fact, contribute to improved patient care, according to a to-be-published study, “The Association Between Hospital Outcomes and Diagnostic Imaging: Could More Be Better?” by David W. Lee, PhD, and David Foster, PhD.
David Lee, PhDThe study’s results indicate that the risk-adjusted probability of receiving an imaging service has a statistically significant inverse correlation with risk-adjusted mortality for all four modalities studied (CT, MRI, ultrasound, and radiography). Utilization of imaging services was correlated with reductions in costs and lengths of stay, though none of the correlations between imaging and cost achieved statistical significance. The authors hope that this work inspires others to investigate further by approaching the issue from different angles. Lee, senior director of health economics and outcomes research for GE Healthcare, Waukesha, Wisconsin, says, “If anything, our study highlights the need for further assessment of whether and how diagnostic imaging shapes key clinical, operational, and economic outcomes in the inpatient setting.” The study has been accepted for publication by a peer-reviewed journal, Lee reports. Sparked by Doubts Lee says that he became interested in conducting this study because of the rising doubts about the value of imaging coming from government and the insurance industry. “Between 2000 and 2006, the use of advanced imaging technologies—such as CT, MRI, PET, and SPECT—grew at more than twice the rate for physician services overall per Medicare beneficiary,” he reports. “As we all know, this spurred Congress to reduce payments significantly for diagnostic imaging services. At the same time, this utilization trend prompted payors and policymakers to begin questioning whether more diagnostic imaging is at all associated with better health outcomes.” Lee notes that the foes of upward utilization trends have been bolstered in their assumptions by dint of a body of literature (no single study in particular) suggesting that when it comes to health care resources in general and imaging specifically, delivery of more does not necessarily result in better outcomes. “Our study was an attempt to focus in on diagnostic imaging in the inpatient setting and see whether that’s true or whether the opposite is the case,” Lee says. Testing the Hypothesis The study by Lee and Foster tests whether more is better in diagnostic imaging. It does this by examining the association between the utilization of inpatient diagnostic imaging services and three key hospital outcome measures: mortality, length of stay, and cost. “We used patient-level data to construct a pair of hospital-specific, risk-adjusted imaging-utilization measures for CT, MRI, ultrasound, and radiography: a binary indicator of whether the patient received the service and an estimate of the mean number of services received,” Lee explains. “Both were adjusted for the patients’ demographic characteristics, severity level, and discharge disposition of death, as well as for hospital characteristics.” The necessary data were culled from the Thomson Reuters 2007 Hospital Drug Database, which covers 1.1 million patients treated at 102 US hospitals. “Nearly 54% of these patients were female, and almost 33% were between the ages of 45 and 64, with another 29.9% between 70 and 84 years old,” Lee notes. “The hospitals ranged from fewer than 200 beds to more than 500 beds. Some were teaching institutions; many were community hospitals. Most were located in the South, but there was a nice mix of urban and rural hospitals.” Lee and Foster chose to sift through the Thompson Reuters database instead of visiting and surveying a cohort of hospitals for reasons of efficiency. “This database—which was available to us in electronic form—offered very detailed information on basically every intervention and service delivered to the 1.1 million patients during each stay,” Lee says. “Included among those interventions and services were the imaging procedures the patients received. Also provided was information about the outcome of care, and how much that care cost.” Ensuring Soundness So that the findings of their investigation would withstand scrutiny, Lee and Foster sought to design the study in such a way that pertinent factors and variables (from one hospital to the next) affecting both the utilization of imaging and patient outcomes would be taken into account. They also took steps to adjust for mortality rates among the hospitals. “Just to illustrate, take an institution like Mayo Clinic,” Lee begins. “You would expect the mortality rate there to be significantly higher than that of, say, a typical community hospital. It would be higher because, of course, Mayo Clinic sees the sickest of the sick. To place Mayo Clinic in the same cohort as a local community hospital, however, it would first be necessary to adjust for the differences in the mortality rates by considering the differences in patient characteristics at Mayo Clinic versus a community hospital. In our study, these and other variances were addressed by building in a number of standard, well-accepted risk-control methods.” Another challenge was finding a way to measure imaging utilization. “We took two approaches to this one,” Lee says. “First, we identified which patients received imaging procedures and which didn’t. Second, for those who received imaging, we counted the number of imaging studies delivered. This required use of sophisticated statistical techniques in order to adjust for compounding factors.” The study indicates that the likelihood of a patient undergoing an imaging service in any of the four modalities had a statistically significant inverse correlation with mortality. The correlations were –0.1964 for MRI, –0.2096 for radiography, –0.2245 for CT, and –0.2397 for ultrasound (see table).
Table. Risk-adjusted Correlations Between Imaging Use and Outcome MeasuresStill Vulnerable Despite the effort to ensure a rigorous study, the work remains vulnerable to criticism. Lee knows this and expects to hear plenty of voices faulting the study’s findings. “One line of attack I’m expecting is that the data come from only a relatively small number of hospitals, and this is true,” he says. “Critics will also say that, while we did as good a job as could be expected in controlling for the factors we identified, there were other factors we didn’t take into account. For example, although we don’t know this to be true, it could be that hospitals invest in imaging not directly to provide better outcomes but to attract higher-quality health care providers.” Another slam against one of the study’s conclusions could be that it failed to show imaging having a dramatic effect on mortality, lengths of stay, and costs (although, in fairness, the correlation between imaging utilization and good outcomes was demonstrated to be statistically significant). “These are all legitimate and perfectly reasonable criticisms,” Lee concedes, “but it should be noted that the literature advancing the position that more imaging is not better employs the exact same research strategies and techniques as does our study, which means the same criticisms directed against ours can be turned around and directed right back at theirs.” Equipped to Challenge When all is said and done, Lee says, the study’s findings represent good news for the imaging community. “Those who can appreciate our work the most will be imaging’s legislative, regulatory, and private-sector advocates,” he says. “Even radiologists, physicians, and administrators with day-to-day clinical decisions to make, however, will welcome the reassurances we offer that they’re doing the right thing for their patients by ordering imaging tests appropriately.” The ultimate gratification will come, for Lee, from knowing that the study better equips the friends of imaging to challenge the current conventional wisdom about imaging’s worth. He asks, “Is there inappropriate imaging? Sure, but by and large, as this study affirms, the vast majority of imaging is appropriate. I don’t know how helpful this study will be in removing the bull’s-eye from radiology. At the very least, it should encourage people to ask whether that bull’s-eye belongs there in the first place.” Rich Smith is a contributing writer for ImagingBiz.com.
David Lee, PhD
Table. Risk-adjusted Correlations Between Imaging Use and Outcome Measures