Winning Abstracts from the 2014 Medical Student Abstract Competition: Association Between Outpatient "No-shows" and Subsequent Clinical Outcomes
First Author: Andrew Hwang, Tufts University School of Medicine
INTRODUCTION: To improve care and achieve performance targets, healthcare systems focus on patients at high risk for not achieving preventive cancer screening and chronic disease management goals. We hypothesized that patients with a higher propensity for missed outpatient appointments, or "no-shows," will have lower colorectal cancer (CRC) screening and low-density lipoprotein (LDL) goal attainment over the following year.
METHODS: We calculated the "no-show propensity factor" (NSPF) for 142,173 patients seen in an academic primary care practice network from 2007-2009 using 5 years (2005-2009) of outpatient appointment data. NSPF transforms a patient's count of arrivals and no-shows into a score that represents appointment adherence, correcting for patients with few appointments to avoid undue influence on the model. We divided patients into two groups: patients in the highest quartile and patients in the lower 3 quartiles of NSPF, and assessed CRC screening completion among patients aged 52-75 years without prior colectomy and LDL goal (LDL<100mg/dl) attainment in patients with diabetes or vascular disease at one year follow-up (2010).
We evaluated the association between NSPF and our two binary outcomes using logistic regression models adjusting for number of visits. Using a "change-in-estimate" strategy, we included patient (age, gender, race, language, insurance, neighborhood median household income, Charlson comorbidity index, and diagnosis of depression or alcoholism) and provider factors (gender, years in practice, and whether the patient received primary care at a community health center) in the models if they altered the odds ratio by >5%. We also ran models adjusting for all variables.
RESULTS: Compared to patients in the lower 3 quartiles for NSPF, patients in the highest quartile were significantly (P<0.0001 for all) younger (45 vs. 52 years) and more likely to be male (44% vs. 42), non-white (39% vs. 18%), non-English speaking (18% vs. 7%), uninsured (7% vs. 4%), live in neighborhoods with low median household income ($50,518 vs. $63,736), receive primary care at a community health center (15% vs. 7%), and have a diagnosis of depression (20% vs. 11%) or alcoholism (3% vs. 0.9%). Compared to eligible patients in the lower 3 quartiles for NSPF, eligible patients in the highest quartile were significantly more likely to have not completed CRC screening (adjusted OR 1.72 [1.63-1.81] adjusting for number of visits and neighborhood median household income) or have an above goal LDL (adjusted OR 1.72 [1.60-1.85] adjusting for number of visits and age). Inclusion of all variables in the models did not change the significance or direction of the effect, and the effect changed by less than 7% for both outcomes.
CONCLUSIONS: NSPF was a strong, independent predictor of subsequent colonoscopy completion and LDL goal attainment. Using NSPF may help healthcare systems identify patients at increased risk for non-adherence to recommended care for targeted interventions to improve care and achieve performance targets.