Date Published: April 26, 2019
Publisher: Public Library of Science
Author(s): Sehee Kim, Fan Wu, Claudia Dahlerus, Deanna Chyn, Yi Li, Joseph M. Messana, Nayu Ikeda.
To assure and improve the quality and safety of care provided by dialysis facilities, federal oversight has been conducted through periodic survey assessment. However, with the growing number of individuals living with ESRD and dialysis facilities, state survey agencies have faced challenges in time and resources to complete survey activities. Therefore, the survey process (‘Basic Survey’ used prior to 2013) was redesigned in order to develop a more efficient process (‘Core Survey’ newly implemented since 2013). The purpose of this analysis was to evaluate and compare dialysis facility survey outcomes between the Core and Basic Survey processes, using a causal inference technique. The survey outcomes included condition-level citations, total citations (condition- and standard-level), and citation rate per survey-hour.
For comparisons of non-randomly assigned survey types, propensity score matching was used. Data were drawn from CMS’ Quality Improvement Evaluation System (QIES) database from January 1, 2013 through July 31, 2014. Covariates available included survey type, facility characteristics (state, urban, practices catheter reuse, dialysis modalities offered, number of patients, mortality, hospitalization, infection) and survey-related characteristics (number of surveyors, time since last survey).
Compared to the Basic Survey, the Core Survey identified 10% more total citations (P = 0.001) and identified condition-level citations more frequently, although the latter finding did not reach statistical significance. These findings suggest an increase of 10% in citation rate (i.e. ratio between citations and survey time) with the Core survey process (P = 0.002).
Greater efficiency has implications for attenuating the time-intensive burden of the state survey process, and improving the safety and quality of care provided by dialysis facilities.
There were 2,571 facilities surveyed between January 2013 and July 2014. We excluded one facility in New York due to incomplete citation counts and 185 facilities due to missing important facility-level covariates that were used in the matching method. The final sample was N = 2,385 facilities, of which 1,275 (53%) underwent Basic Surveys and 1,110 (47%) were surveyed with the Core Survey. In Table 1, we observed systematic differences in facility and survey characteristics between the two survey processes in the original unmatched sample. In particular, the Core Survey was more likely used in rural areas, when the survey team size was as small as one, and for facilities where 3.5 years or more had elapsed since their last survey. These factors could potentially lead to confounding in these results. Therefore, we took these into account in our modeling by utilizing the propensity score matching method, a statistical method specifically designed to reduce confounding biases in our setting.
Federal regulatory oversight provides the policy context for our study results. Congressional investigation of dialysis facility oversight led to multiple calls to increase the frequency, consistency and breadth of the dialysis survey process. Our results suggest the Core Survey in the initial implementation period yielded greater efficiency, as measured by the rate of identifying citations (regulatory violations) per on-site hour compared to the Basic Survey. In our propensity score matched analyses, the Core Survey process identified 10% more total citations and was 10% more efficient than the Basic Survey. While the Core Survey resulted in identifying more condition-level citations, the absence of statistical significance for this result could be attributed to the overall observed lower event counts for these severe violations.