Constantin Shuster, MD, Andrew Hurlburt, MD, Penny Tam, MD, John A. Staples, MD, MPH

Unplanned hospital readmissions in British Columbia

Reducing the rate of unplanned hospital readmissions can address associated patient discontent, increased health care costs, and increased risks for morbidity and mortality.

ABSTRACT: Rates of unplanned hospital readmissions. No single n 2009 a landmark study found hospital readmissions are publicly intervention has been successful that nearly 20% of US Medicare reported in Canada and often inter- in reducing unplanned readmis- Ibeneficiaries were readmitted to preted as a marker of health care sion thus far; multiple-component hospital within 30 days, prompting system performance. In 2016 Brit- interventions have shown promise, hospital readmissions to become a ish Columbia’s 30-day risk-adjusted but their success has proven dif- major focus of health care quality im- readmission rate of 9.7% was higher ficult to replicate. Clinicians and provement efforts.1 Subsequent rec- than the national average of 9.1%. administrators aiming to reduce ognition of wide regional variability This is regrettable because read- unplanned readmissions should in readmission rates suggested that a missions are associated with pa- consider tracking local readmis- proportion of hospital readmissions tient discontent, increased health sion rates, implementing context- might be preventable if a focused ef- care costs, and increased risks of appropriate interventions, and us- fort was made to improve hospital morbidity and mortality. The fact ing risk-prediction models to iden- and community care.1,2 A number of that readmissions affect many Ca- tify and target patients at the highest organizations in Canada, the United nadian patients and cost more than risk of readmission. Given the poor Kingdom, and the United States now $1.8 billion per year should moti- outcomes and increased costs asso- recognize a high rate of unplanned vate clinicians, hospitals, and health ciated with hospital readmissions,­ a hospital readmission as a marker authorities to institute programs to concerted effort should be made to monitor and prevent unplanned address this issue. Dr Shuster is a senior resident in the Di- vision of General Internal Medicine in the Department of Medicine at the University of British Columbia. Dr Hurlburt is a senior resident in the Division of General Internal Medicine at UBC. Dr Tam is a clinical as- sistant professor in the Division of General Internal Medicine at UBC. Dr Staples is a clinical assistant professor in the Division of General Internal Medicine at UBC and a researcher at the Centre for Clinical Epide- miology and Evaluation based at Vancouver This article has been peer reviewed. General Hospital.

bc medical journal vol. 60 no. 5, june 2018 bcmj.org 263 Unplanned hospital readmissions in British Columbia

of suboptimal health care system performance.3 According to the Canadian Insti- tute for Health Information (CIHI), unplanned hospital readmissions affect almost 200 000 Canadians annually.4 Unfortunately, British Co- lumbia’s 2016 risk-adjusted 30-day readmission rate of 9.7% was signifi- cantly higher than the national average 9.7 of 9.1% ( Figure 1 ).5 The readmission rates in Saskatchewan (9.7%) and On- 9.7 tario (9.2%) were also higher than the national average. Manitoba (8.7%), Quebec (8.6%), Nova Scotia (8.5%), and New Brunswick (8.8%) had rates that were significantly lower than the national average. As well, in 2016 Vancouver Coastal Health had a read- Figure 1. Risk-adjusted 30-day readmission rate (%) by province and territory. Based on data mission rate of 9.8%, which was the for 2015–2016 obtained from CIHI.5 second highest among BC’s five re- gional health authorities and exceeded both national and provincial averages creased health care costs, and increas­ pitalization is often greater than the ( Figure 2 ). These comparisons high- ed risks for morbidity and mortality. first ($10 404 versus $7287 for medical light an opportunity to improve the Patient dissatisfaction may arise from patients).4 Hospital readmissions may performance of BC’s health system. the perception that the readmission be complicated by iatrogenic infec- was preventable.6,7 Hospital readmis- tion, venous thromboembolism, drug Why do readmissions sions cost Canadian taxpayers over reactions, falls, and pressure ulcers.8 matter? $1.8 billion per year, which represents Large cohort studies have found the Unplanned hospital readmissions are 11% of annual inpatient costs.4 More- mortality rate after a hospital readmis- associated with patient discontent, in- over, the average cost of a second hos- sion to be 19% at 30 days and 39% at 1 year; the latter represents a threefold increase in risk for patients who were readmitted compared with patients who remained in the community after Vancouver Coastal Health hospital discharge.9,10 Providence Health Care*

Fraser Health How are readmission rates tracked? Island Health Hospital readmission rates are calcu- Interior Health lated by determining the proportion of discharged patients who are readmit- Northern Health ted within a designated time frame. 8.0 9.0 10.0 11.0 A 30-day time frame is usually used, 30-day readmission rate (%) although there is no clear biological justification for this choice.11-13 Eligi- bility criteria for the numerator and Figure 2. Risk-adjusted 30-day readmission rates for BC regional health authorities compared with national average (dashed line). Based on data for 2015–2016 obtained from CIHI.5 denominator often differ among insti- * Providence Health Care is not a health authority; however, it is a major hospital network within the tutions, making it difficult to compare Lower Mainland with many physicians and administrators to whom this statistic may be relevant. hospitals’ self-reported readmission

264 bc medical journal vol. 60 no. 5, june 2018 bcmj.org Unplanned hospital readmissions in British Columbia

rates. For example, planned readmis- sions (e.g., for elective surgery) are 20 frequently excluded from the numer- 18 ator, but only some hospitals exclude 16 psychiatric and palliative discharges 14 from the denominator. Hospital-based 12 BC average (9.7%) tracking programs also often fail to 10 consider the 20% of readmissions 8 that are known to occur at a different 6 hospital.9 Standardized reporting by 4

30-day readmission rate (%) 2 CIHI overcomes many of these chal- 0 lenges and facilitates equal compari- Obstetric Pediatric Medical Surgical sons between hospitals and regions ( 19 years) by accounting for site-specific differ- ences in patient age and comorbidity burden. Figure 3. Program-specific 30-day readmission rates for British Columbia. Based on data for 2015–2016 obtained from CIHI.5 Who is at risk? Patient risk factors for unplanned controversial.14,15 The full impact of 17.3%).21 However, recent analyses hospital readmission include male length of stay on risk of readmission found that the introduction of the sex, advanced age, increased comor- is not fully understood as other stud- HRRP was associated with a 30-day bidity burden, lower socioeconomic ies have found a longer length of stay mortality rate increase after an ad- status, and increased hospitalizations to be associated with a higher risk for mission for heart failure (from 7.2% within the last 6 months.4 Patients readmission.16 to 8.6%).22 Further debate over the with medical admissions are at high- merits of this program is inevitable. est risk for readmission ( Figure 3 ). Are readmissions Local researchers believe that imple- About 20% of patients initially admit- preventable? mentation of an HRRP-like policy in ted for chronic obstructive pulmonary About 25% of unplanned hospital re- BC is unlikely, in part because global disease (COPD) or heart failure are admissions are retrospectively deter- hospital budgets make such dis­ readmitted within 30 days. Among mined to be preventable, but reliably incentives less effective.23 surgical patients, those undergoing effective and focused interventions colostomy or enterostomy are at high- to prevent them remain elusive.11,17,18 How can readmissions est risk for readmission. The main in- Multiple-component interventions, be addressed? dependent readmission risk factor in specifically where at least three strat- Clinicians and administrators may any patient is having been hospital- egies are used to reduce readmis- consider tracking the local readmis- ized twice or more in the 6 months sions, have shown promise but have sion rate, implementing context- before the index admission. Hospi- been difficult to replicate.19,20 The appropriate interventions, and refin- tal-specific risk factors for readmis- largest and most effective readmis- ing their approach with sequential sion are small patient volume (fewer sion reduction effort to date is the plan-do-study-act quality improve- than 2000 weighted cases annually) ongoing Hospital Readmissions Re- ment cycles.24 Risk prediction mod- and rural location. Hospitals with a duction Program (HRRP) in the US. els such as the LACE index and the longer average length of stay have Through the HRRP policy, hospitals HOSPITAL score can be used to help lower risk-adjusted readmission rates. with higher-than-expected condition- identify patients at the highest risk of On average, discharging a patient at specific 30-day readmission rates for readmission.25-27 Frameworks for de- least 1 day earlier than the national US Medicare patients are financially veloping readmission risk-reduction expected length of stay increases the penalized. This has resulted in signifi- interventions are available from the relative risk of readmission by around cant reductions in the 30-day readmis- Institute for Healthcare Improve- 40%.4 The cumulative influence that sion rate for both targeted conditions ment’s STate Action on Avoidable Re- these competing forces have on cost (from 24.1% to 22.5%) and for non- hospitalizations (STAAR) program to the health care system remains targeted conditions (from 17.8% to and from the Care Transitions Pro-

bc medical journal vol. 60 no. 5, june 2018 bcmj.org 265 Unplanned hospital readmissions in British Columbia

Vancouver Coastal Health Island Health

Vancouver General and UBC Victoria General and Royal Jubilee Providence Health Care Nanaimo Reigional General Lions Gate Cowichan District Richmond St. Joseph's General [Comox] West Coast General St. Mary's [Sechelt] Campbell River and District General Powell River Saanich Peninsula Squamish Lady Minto Gulf Islands Bella Coola Port McNeill and District R.W. Large Memorial Port Hardy 0 5 10 15 20 Tofino General 30-day readmission rate (%) 0 5 10 15 20 30-day readmission rate (%)

Fraser Health Interior Health

Surrey Memorial Kelowna General Royal Columbian Royal Inland Abbotsford Regional Vernon Jubilee Burnaby Penticton Regional Langley Memorial East Kootenay Regional Kootenay Boundary Regional Peace Arch Shuswap Lake General Eagle Ridge Cariboo Memorial Chilliwack General Kootenay Lake Ridge Meadows South Okanagan General Delta Creston Valley Mission Memorial Elk Valley 100 Mile District General Fraser Canyon [Hope] Invermere and District 0 5 10 15 20 Nicola Valley Boundary 30-day readmission rate (%) Golden and District General Queen Victoria Dr. Helmcken Memorial Princeton General Lillooet Northern Health Arrow Lakes 0 5 10 15 20 Univ. Hospital of Northern B.C. 30-day readmission rate (%) G.R. Baker Memorial Fort St. John Mills Memorial Prince Rupert Dawson Creek and District Bulkley Valley District Kitimat Lakes District St. John Fort Nelson General Stuart Lake Wrinch Memorial Queen Charlotte Islands General Chetwynd Mackenzie and District Northern Haida Gwaii McBride and District 0 5 10 15 20 30-day readmission rate (%)

Figure 4. Risk-adjusted 30-day readmission rates for BC hospitals compared with regional health authority averages (dashed lines). Based on data for 2015–2016 obtained from CIHI.5

266 bc medical journal vol. 60 no. 5, june 2018 bcmj.org Unplanned hospital readmissions in British Columbia

gram.28 CIHI data can be used to S, et al. Readmissions in the era of patient deemed avoidable: A systematic review. continue comparing progress among engagement. JAMA Intern Med 2014; CMAJ 2011;183:E391-402. provinces, health authorities, and hos- 174:1870-1872. 19. Bradley EH, Sipsma H, Horwitz LI, et al. pitals ( Figure 4 ).5 8. Wang Y, Eldridge N, Metersky ML, et al. Hospital strategy uptake and reductions in Unplanned hospital readmissions National trends in patient safety for four unplanned readmission rates for patients are a major burden on health care sys- common conditions, 2005-2011. N Engl with heart failure: A prospective study. J tems in BC and nationwide. Given the J Med 2014;370:341-351. Gen Intern Med 2015;30:605-611. poor outcomes and high costs associ- 9. Staples JA, Thiruchelvam D, Redelmeier 20. Leppin AL, Gionfriddo MR, Kessler M, et ated with readmissions, a concerted DA. Site of hospital readmission and mor- al. Preventing 30-day hospital readmis- effort should be made to address this tality: A population-based retrospective sions: A systematic review and meta-anal- issue with the help of those working cohort study. CMAJ Open 2014;2: ysis of randomized trials. JAMA Intern at all levels of the health care system, E77-85. Med 2014;174:1095-1107. including clinicians, hospital admin- 10. Lum HD, Studenski SA, Degenholz HB, 21. Zuckerman RB, Sheingold SH, Orav EJ, et istrators, and policymakers. Hardy SE. Early hospital readmission is al. Readmissions, Observation, and the a predictor of one-year mortality in com- Hospital Readmissions Reduction Pro- Competing interests munity-dwelling older Medicare benefi- gram. N Engl J Med 2016;374:1543-1551. None declared. ciaries. J Gen Intern Med 2012;27:1467- 22. Gupta A, Allen LA, Bhatt DL, et al. Asso- 1474. ciation of the Hospital Readmissions Re- References 11. van Walraven C, Jennings A, Taljaard M, duction Program implementation with 1. Jencks SF, Williams MV, Coleman EA. Re- et al. Incidence of potentially avoidable ur- readmission and mortality outcomes in hospitalizations among patients in the gent readmissions and their relation to all- heart failure. JAMA Cardiol 2018;3:44-53. Medicare fee-for-service program. N Engl cause urgent readmissions. CMAJ 2011; 23. Hellsten E, Liu G, Yue E, et al. Improving J Med 2009;360:1418-1428. 183:E1067-1072. hospital quality through payment reforms: 2. The Dartmouth Atlas of Health Care. Un- 12. Graham KL, Wilker EH, Howell MD, et al. A policy impact analysis in British Colum- derstanding of the efficiency and effec- Differences between early and late read- bia. Healthc Manage Forum 2016;29: tiveness of the health care system. missions among patients: A cohort study. 33-38. Accessed 2 April 2018. www.dartmouth Ann Intern Med 2015;162:741-749. 24. Li J, Williams MV. Care transitions: It’s the atlas.org. 13. Vaduganathan M, Bonow RO, Gheorghi- how, not just the what. J Gen Intern Med 3. Anderson GF, Steinberg EP. Hospital read- ade M. Thirty-day readmissions: The clock 2015;30:539-540. missions in the Medicare population. N is ticking. JAMA 2013;309:345-346. 25. Kansagara D, Englander H, Salanitro A, et Engl J Med 1984;311:1349-1353. 14. Kaboli PJ, Go JT, Hockenberry J, et al. As- al. Risk prediction models for hospital re- 4. Canadian Institute for Health Information. sociations between reduced hospital admission: A systematic review. JAMA All-cause readmission to acute care and length of stay and 30-day readmission rate 2011;306:1688-1698. return to the emergency department. Ot- and mortality: 14-year experience in 129 26. van Walraven C, Dhalla IA, Bell C, et al. tawa: CIHI; 2012. Accessed 2 April 2018. Veterans Affairs hospitals. Ann Intern Derivation and validation of an index to https://secure.cihi.ca/free_products/ Med 2012;157:837-845. predict early death or unplanned readmis- Readmission_to_acutecare_en.pdf. 15. Carey K. Measuring the hospital length of sion after discharge from hospital to the 5. Canadian Institute for Health Information. stay/readmission cost trade-off under a community. CMAJ 2010;182:551-557. Your health system: All patients readmit- bundled payment mechanism. Health 27. Donze J, Aujesky D, Williams D, Schnip- ted to hospital. Accessed 2 April 2018. Econ 2015;24:790-802. per JL. Potentially avoidable 30-day hospi- http://yourhealthsystem.cihi.ca/hsp/ 16. Westert GP, Lagoe RJ, Keskimaki I, et al. tal readmissions in medical patients: Der- inbrief?lang=en#!/indicators/006/all An international study of hospital readmis- ivation and validation of a prediction -patients-readmitted-to-hospital/;mapC1; sions and related utilization in Europe and model. JAMA Intern Med 2013;173: mapLevel2;provinceC9001;trend(C1). the USA. Health Policy 2002;61:269-278. 632-638. 6. Boulding W, Glickman SW, Manary MP, et 17. van Walraven C, Jennings A, Forster AJ. 28. The Care Transitions Program. Tools and al. Relationship between patient satisfac- A meta-analysis of hospital 30-day avoid- resources. Accessed 2 April 2018. https:// tion with inpatient care and hospital read- able readmission rates. J Eval Clin Pract caretransitions.org/tools-and-resources. mission within 30 days. Am J Manag Care 2012;18:1211-1218. 2011;17:41-48. 18. van Walraven C, Bennett C, Jennings A, 7. Howard-Anderson J, Lonowski S, Vangala et al. Proportion of hospital readmissions

bc medical journal vol. 60 no. 5, june 2018 bcmj.org 267