Language Concordant Home Care Visits Reduce 30 Day Readmissions in - PowerPoint PPT Presentation

Language Concordant Home Care Visits Reduce 30 Day Readmissions in Limited English Proficiency Patients Allison Squires, PhD, RN, FAAN Sarah Miner, PhD, RN Chenjuan Ma, PhD Penny Feldman, PhD (VNSNY) Simon Jones, PhD (NYU SOM) 2018

Background: Qualifying for Home Health Care • Diabetes: newly-diagnosed, • A newly-diagnosed, change, uncontrolled, unable to self- or decline in condition manage, or requiring education • Therapy Needs: Physical, • Frequent hospitalizations or Occupational, Nutritional, and emergency room use Speech • Unsteady gait, falls, or fractures, • Difficulty leaving the home or joint replacement recovery is ‘homebound’ • Medication management needs • Teaching Needs: disease • Unhealed wounds requiring process, injections, I.V. management therapy, etc. Source: https://www.medicare.gov/coverage/home-health-services.html#1334 2 6/20/2018

Background: Home Health Care in the US Source: 2018 MedPac Report 3 6/20/2018

Home Care Has the Potential to Reduce 30 Day Readmissions • Sources of readmissions are not well differentiated in research (Burke & Ibrahim 2018) • Home care services can reduce modifiable readmission risks, but the topic is understudied (Ma et al 2017) • Continuity of care (i.e. same nurse seeing the patient) improves home care outcomes (David & Kim 2018) • Communication failures between nurses & physicians contribute to readmission risk (Pesko et al 2018) 4 6/20/2018

Background – Limited English Proficiency Patients • One in 5 US households speaks a language other than English at home (Ryan 2013) • Limited English Proficiency (LEP) individuals do not speak English well enough to safely and effectively communicate with their health care provider • LEP patients are accessing home care services in growing numbers annually

Readmission & LEP Patients in Home Care • What is known • What we need to know o Sources of health o No studies have specifically disparities in home care examined LEP readmission are not well understood (Davitt risk from home care et al 2015; Narayan & Scafide 2017) o Studies do not differentiate o LEP status is a known risk risk by language preference factor for hospital readmission, with risk rates o Interpreting modality’s varying between 15-25% effects on readmission rates (Karliner et al 2010) not known o Discharge instructions rarely come in the language of the patient (Karliner et al 2012) 6 6/20/2018

Purpose • To compare the 30-day all cause readmission among LEP patients recently discharged from the hospital who receive home care services from a language concordant nurse to those who do not. • Parent Study: A Mixed Methods Analysis of Post-Acute Outcomes in Limited English Proficiency Home Care Patients (AHRQ R01 HS023593) 7 6/20/2018

Data Sources • Dates: January 1, 2010 to March 15, 2015 • Outcomes Assessment Information System (OASIS) records • Visiting Nurse Service of New York (VNSNY) Administrative Data • VNSNY human resources data for staff reported language skills 8 6/20/2018

Methods • Sample of 90,421(including English speakers), with 19,654 LEP only • Inverse probability weighting with a marginal structural model conducted • Matched for continuity of care factors (e.g. did the same person care for the patient throughout the episode?) • As a robustness check, multiple regression model to look at predictive factors 9 6/20/2018

Question 1: Does readmission rate vary by language preference? 0.7 • Matching 0.6 variables before 0.5 0.4 SMD and after 0.3 weighting 0.2 0.1 0 • Treatment = English preferred language Unweighted Weighted Limit

Results: Continuity of Care (COC) by Language Preference (Unweighted) Better Worse Kruskal-Wallis rank sum test p < 0.001

Adjusting for COC is Important in Home Care When Assessing Readmission Rates & Risk ATE = Average Treatment Effect 12 6/20/2018

Results: Language Preference Other Than English Affects Readmission Rates from Home Care Readmission rates by patient’s language preference Readmission N of Patients Readmitted 95% CI Rate Overall 87,561 17,097 19.5% 19.3% - 19.8% English 66,079 12,567 19.0% 18.7% - 19.3% Spanish 17,656 3,824 21.7% 21.1% - 22.3% Chinese 1,721 278 16.2% 14.4% - 17.9% Korean 341 56 16.4% 12.6% - 20.5% Russian 1,764 372 21.1% 19.2% - 23.0% Kruskal-Wallis chi-squared = 37.957, p-value = 7.232e-10 13 6/20/2018

Question 2: Does a language concordant nurse reduce the risk for readmission? • It's complicated. • Explanations from the qualitative data (73 LEP patients, 35 home care staff): o Patients request LC providers and VNSNY tries to accommodate o Some, but not all, managers make staff assignments based on language skills o Staff who speak other languages frequently take on extra assignments outside their service area 14 6/20/2018

Results from Adjusted Data ATE language concordance on of Re-hospitalization 25% 50% 75% 100% 2.00% 1.00% 0.00% -1.00% ATE -2.00% -3.00% -4.00% -5.00% -6.00% Language Concordance 15 6/20/2018

Odds of Rehospitalization Depends on a Combination of COC and LC Nurses Odds Ratio of Rehospitalization 1.04 1.02 1 0.98 0.96 0.94 0.92 0.9 0.88 LowCOC, Low LngC Reference Low COC, High LangC (n=547) High COC, Low Lang C (n=7742) High COC, High Lang C (n=2183) (n=9182) 16 6/20/2018

Summary of Results • If we control for COC, LC visits need to reach 90% to reduce readmission. • Logistic regression with 4 groups says: o Just language concordance has no effect on readmission rate o Having COC has an effect on readmission rate o Having both COC and LC is most effective in reducing readmission rate • Risk for reverse causality due to how managers assign staff (confirmed by qual data) 17 6/20/2018

Limitations • Single site, urban study, long organizational experience with LEP patients • Assumption that there are no unobserved confounders, but plausible ones do exist • Challenging disentangle the impact of COC and LC • Data on Spanish speakers may largely apply to Puerto Rican and Dominican populations due to NYC demographics • “Chinese” classification in VNSNY system does not differentiate by dialect, so outcomes between Mandarin, Cantonese, and other dialect speakers not differentiable 18 6/20/2018

Discussion • Readmission risk from home care is differentiated by patient language preference • Continuity of care combined with language concordant visits may offer best intervention for reducing readmission risk in LEP patients o But US workforce demographics do not match demand for LC services • Based on the data, does telephone interpretation really appear to be effective in home health care? 19 6/20/2018

Acknowledgements • NYU & Other Study • VNSNY Team Members o Melissa Trachtenberg o Eva Liang o Penny Feldman o Gavin Arneson o Yolanda Barron Vaya o Maichou Lor o Sridevi Sridharan o Gary Yu o VNSNY staff o Research Assistants o Consultant: Liz Jacobs, • Funding: MD, MPAA o AHRQ R01 HS023593

Allison Squires, PhD, RN, FAAN aps6@nyu.edu 21 6/20/2018