Selecting high priority quality measures for breast cancer quality improvement

Abstract

Background: Although many quality measures have been created, there is no consensus regarding which are the most important. We sought to develop a simple, explicit strategy for prioritizing breast cancer quality measures based on their potential to highlight areas where quality improvement efforts could most impact a population.

Methods: Using performance data for 9019 breast cancer patients treated at 10 National Comprehensive Cancer Network institutions, we assessed concordance relative to 30 reliable, valid breast cancer process-based treatment measures. We identified 4 attributes that indicated there was room for improvement and characterized the extent of burden imposed by failing to follow each measure: number of nonconcordant patients, concordance across all institutions, highest concordance at any 1 institution, and magnitude of benefit associated with concordant care. For each measure, we used data from the concordance analyses to derive the first 3 attributes and surveyed expert breast cancer physicians to estimate the fourth. A simple algorithm incorporated these attributes and produced a final score for each measure; these scores were used to rank the measures.

Results: We successfully prioritized quality measures using explicit, objective methods and actual performance data. The number of nonconcordant patients had the greatest influence on the rankings. The highest-ranking measures recommended chemotherapy and hormone therapy for hormone-receptor positive tumors and radiation therapy after breast-conserving surgery.

Conclusions: This simple, explicit approach is a significant departure from methods used previously, and effectively identifies breast cancer quality measures that have broad clinical relevance. Systematically prioritizing quality measures could increase the efficiency and efficacy of quality improvement efforts and substantially improve outcomes.

Figure 1. Algorithm Used to Generate Scores

Four attributes were incorporated into an algorithm that was used to generate scores for 30 quality measures. Quality measures were prioritized as targets for quality improvement based on their final scores. Three attributes – the number of non-concordant patients, highest concordance at any one institution, and overall concordance across all institutions – were derived from analyzing the treatments provided to 9019 women with breast cancer relative to the recommendations made by the National Comprehensive Cancer Network. One attribute – the fractional magnitude-of-benefit estimate – was derived by asking a panel of expert breast cancer clinicians to estimate the benefit experienced by patients who received a recommended treatment compared to patients who received a common non-concordant treatment.

Figure 2. Five-year disease-free survival and quality-of-life scores versus magnitude-of-benefit estimates for each treatment recommendation

Recommendations for treatments (closed diamonds) are presented separately from recommendations against treatments (open squares). The estimated improvement in disease-free survival experienced by a population of patients receiving the recommended instead of a non-concordant treatment was reported as the percent absolute benefit at five years. The estimated difference in quality of life was reported as greatly favors recommended treatment, slightly favors recommended treatment, no difference, slightly favors non-concordant treatment, or greatly favors non-concordant treatment. Trend lines were generated using simple linear regression. Spearman correlation coefficients (r values) are presented with their corresponding P values.

Figure 3. Comparison of values for each quality measure

Results are presented in parallel for each quality measure. Measures are listed in order from highest to lowest final score. The figure includes the measure number (as it appears in table 1), the total number of eligible patient encounters, the number of patients who received care that was not concordant with each treatment recommendation, the highest and overall percent concordance values (solid black and grey bars, respectively), the mean magnitude-of-benefit estimates, and the final scores. The values that appear in the second, third, and fourth graphs were used to calculate the final scores that appear in the fifth graph. Measures 13, 12 and 8 (ranked 28^th, 29^th and 30^th, respectively) had fewer than 10 eligible patients at all centers, so reliable highest-concordance values could not be calculated (and therefore, final scores were not derived for these three measures).