3 ways to cut variability in CT radiation dose

A Pennsylvania hospital network was able to reduce variation in CT radiation dose by taking three key steps, according to research presented at RSNA 2017 and published online December 1 in the Journal of the American College of Radiology.

Researchers implemented a multipronged approach to minimize the variation involved in selecting an appropriate radiation dose for CT exams throughout Einstein Healthcare Network in Philadelphia. Their intervention lowered radiation dose variability by 49% in chest CT exams and by even more in exams of the head (67%) and abdomen and pelvis (60%) across the entire network (JACR, December 1, 2017).

"The standardization of protocols and the ability to use dose-modulation technology adequately allowed us to significantly decrease our variability," Dr. Ryan Lee told attendees at RSNA 2017. "Neither of these two major things would have occurred without the overarching authority of the committee that we created."

A multifaceted approach

Dr. Ryan Lee from Einstein Healthcare Network.

In recent years, healthcare facilities worldwide have been reducing the radiation dose they use for CT exams by applying iterative reconstruction, establishing diagnostic reference levels, and other various methods. Although these techniques have helped drop radiation dose overall, research has also shown that considerable disparities in radiation dose persist around the globe and even within the same health network.

Administering a consistent radiation dose for each CT exam for every patient is a hallmark of high quality in radiation management, Lee said. Unfortunately, reducing dose alone does not necessarily correlate with decreases in variability.

"Mathematically, as we reduce the average of something, we expect the variability and standard deviation to also go down," he said. "In a perfect world, that may be the case. However, in medicine, we all know that things are rarely, if ever, equal."

The numerous factors that account for dose variation include the technologists and radiologists' personal preferences for settings or protocol use, site-specific culture, differing dose-modulation techniques, and diverse types of scanners.

To combat this kind of inconsistency in CT, the researchers developed a multifaceted approach they broadly categorized into three parts:

• Establishing a radiation dose optimization committee. Comprising radiologists, residents, radiation safety officers, medical physicists, and technologists, the committee followed evidence-based medicine to develop a protocol that was implemented at one site and only spread to the entire network if the resulting data proved satisfactory.

  "This committee was important because it allowed overarching authority to implement protocols across the network and also enforce adherence to protocols and policies," Lee said. "It also served as the venue for feedback."

• Standardizing protocols. The committee directed the adoption of universal protocols under the guidance of section chiefs and supervising technologists at each site in the network.

• Implementing dose-modulating scanner software. The software (iPatient, Philips Healthcare) automatically increases tube current and permits slightly more noise as patient size increases -- allowing for a controlled increase in dose as opposed to the exponential increase characteristic of other conventional dose-modulation systems.

Reducing dose variation

To assess the effectiveness of their program, the researchers measured CT radiation dose before and after the program was implemented. They used dose tracking software (Radimetrics, Bayer HealthCare) to collect radiation dose measurements and variation for CT exams before the program was implemented, from January 2013 to July 2014, and then afterward from January 2016 to December 2016. The bulk of the data came from unenhanced CT scans of the head and chest as well as enhanced CT scans of the abdomen and pelvis.

On average, the nearly 50,000 CT scans in the postintervention period had a 39% lower CT dose index volume (CTDI_vol) and 47% less variation, i.e., standard deviation, than the approximately 40,000 CT scans in the preintervention period.

Besides body region, other factors reflected differing changes in dose variability, such as the department where the scanner was used. For example, for a scanner used for inpatients and outpatients, variation in radiation dose declined 42% after the program was implemented. But for a scanner in the emergency department (ED), variation only dropped 10% (p < 0.0001).

In a similar fashion, the standard deviation of exams performed on Philips scanners decreased by 47%, compared with a 9% decrease on GE Healthcare scanners, though again, both scanners were associated with a significantly decreased variance after intervention (p < 0.0001).

Both adult and pediatric populations also displayed statistically significant reductions in radiation dose variation (p < 0.0001): Adults showed a 47% drop in standard deviation, while children showed a 52% reduction.

Promoting dose consistency

To further verify the extent of dose reduction, the researchers juxtaposed their radiation data with the American College of Radiology's (ACR) Dose Index Registry (DIR) from January to June 2016. The Einstein Healthcare Network's median CT radiation dose was considerably lower on average, with a CTDI_vol of 20 for head CT exams, 5 for the chest, and 7 for the abdomen and pelvis, compared with the ACR's 53 for the head, 9 for the chest, and 15 for the abdomen and pelvis.

While reducing radiation is an important goal, just as important in the management of radiation dose for a specific patient is the probability that the person will actually get that low dose, Lee said.

"For an individual patient, we believe that dose variability -- in other words, consistency of radiation dose -- is very important," he said.

A notable limitation of the study was the difficulty in assessing the individual effect of components of the intervention on reducing dose variation, according to the researchers. In addition, comparing data from scanners by different manufacturers is questionable in this case due to newer technology being available for some but not all scanners.

The researchers' findings have encouraged them to expand the scope of their investigation to cover variation in radiation dose for other imaging modalities as well.

"We are currently working on delving further into population types (inpatient, outpatient, ED) to see if we can isolate other factors," Lee told AuntMinnie.com. "On a similar but related note, we are going to start monitoring doses for x-rays and interventional procedures, and we hope to use similar principles to optimize dose and reduce variability in these modalities."