Statistical process control charts for ophthalmology

Abstract

Dear Editor,In this era of evidence-based medicine, ophthalmologistscontinually strive to integrate the most advanced medicaland surgical knowledge into routine clinical practice.However, do we have any real-time methods to statisticallymonitor processes of care in our practices and revealavenues for improvement? Over the past several years, theuse of statistical process control (SPC) charts, a toolkitdeveloped in engineering for process improvement, hasgained popularity in healthcare. Though SPC charts aretypically used in primary care settings, specialties includingcardiology and pulmonology have begun to utilize thesetools to assess both patient outcomes and the effectivenessof quality improvement interventions into clinical processes[1, 2].There are several types of SPC charts. XmR charts andp-charts are frequently used in the healthcare setting [2].These charts characterize the incidence of disease longitu-dinally over a given period of time. They can be used inconjunction with g-charts, which are designed to monitorthe number of days between incidence of disease. Interpre-tation of a g-chart enables the user to predetermine howmuch variation (i.e., the interval days between cases) ispermissible before case frequency becomes excessive, thuscreating a goal to guide quality improvement measures.SPC charts not only monitor incidence over time but canalso be used to calculate points of statistically significantvariation—‘special cause’—in the clinical process [2].Initially, physicians must determine the normal clinicaloutcome variation for a given procedure. Benchmarks fornormal variation can be foundintheliterature.Forexample, assume that at a particular clinic there have beenapproximately 3–4 cases of endophthalmitis each year forthe past 10 years, and that 6,000–8,000 surgeries have beenperformed each year. If one were to track this data over adecade, it would most likely be the case that the data wouldbe in normal (‘common cause’) variation. Normal variationis stable and predictable. However, if there is a ‘specialcause’ signal, it denotes that there has been a change to theprocess or system either above the upper control limit(UCL) or below the lower control limit (LCL). Specialcause indicates that the incidence of cases has exceedednormal variation in a given time (e.g., seven cases peryear). In the case of endophthalmitis occurrence, if thespecial cause signal occurs below the LCL, this lets oneknow that there is something positive occurring in theclinical process and that one must continue to do it in orderto continue such progress. As one can discern, qualitativejudgment must be used to decide on a clinically acceptablelevel of complications when trying to determine if a processis to be continued or if changes to the process arewarranted.There have been only two reports in the ophthalmologicliterature exploring the use of control charts [3, 4]. Both