Emergency department visits during outpatient parenteral antimicrobial therapy: a retrospective cohort study

Emergency department visits during outpatient parenteral antimicrobial therapy: a retrospective... Abstract Objectives Patients receiving outpatient parenteral antimicrobial therapy (OPAT) may require emergency department (ED) visits to manage complications. This study’s purpose was to identify risk factors for ED visits during OPAT and risk factors for hospitalization among patients with ED visits. Methods All OPAT courses initiated between 1 January 2013 and 1 January 2017 at Cleveland Clinic were identified. The first OPAT course per patient was included. For these, ED visits within 30 days were identified. Reasons and risk factors for these visits were sought, as were risk factors for hospitalization among patients visiting the ED. Results Among 8263 patients on OPAT, 381 (4.6%) had at least one ED visit, an additional 1133 (14%) were hospitalized and an additional 50 (0.6%) died, within 30 days. One hundred and ninety-three ED visits (51%) were OPAT related. In a multivariable subdistribution proportional hazards competing risks regression model, prior ED visit (preceding year) was most strongly associated with ED visits (HR 2.29, 95% CI 1.76–2.98, P = 8.1 × 10−10). Eighty-five visits (22%) led to hospitalization. Compared with non-OPAT-related reasons, visits for vascular access complications were associated with lower odds of hospitalization (OR 0.36, 95% CI 0.14–0.83, P = 0.022) and visits for worsening infection with higher odds (OR 18.95, 95% CI 5.50–79.85, P = 1.2 × 10−7). Conclusions Compared with patients without, patients with prior ED visit have a 2.3-fold higher hazard of an ED visit within 30 days of OPAT initiation. Visits for worsening infection are much more likely to result in hospitalization than those for vascular access complications. Introduction Outpatient parenteral antimicrobial therapy (OPAT) has become a widely accepted form of therapy for patients who require intravenous antimicrobial therapy at home or other outpatient setting and it is estimated to be provided to 1 out of 1000 Americans every year.1 It has allowed for shorter hospitalizations, resulting in lower costs of healthcare.1,2 The initial model of OPAT was the visiting nurse model, where parenteral antimicrobials were infused in patients’ homes by visiting nurses. Over the years, many other models have been developed including the infusion centre model and the teach-and-train model.3 The visiting nurse model in the USA has evolved into primarily a self-administration model with periodic (typically once weekly) visiting nurse support. The goal of all these models is to allow safe administration of parenteral antimicrobials outside an inpatient hospital setting. Not every OPAT course concludes uneventfully. Unanticipated events that require attention include treatment failure, antimicrobial adverse events and vascular access complications. Some of these events lead to hospital admission. Previous hospitalization in the preceding 12 months, history of malignant lymphoma, longer planned OPAT duration, higher age, history of drug-resistant organisms, aminoglycoside use and lack of a primary care doctor have been identified as risk factors for hospital admission during OPAT.4,5 OPAT duration, female sex and injection drug use have been identified as risk factors for vascular access complications.6 Some patients on OPAT develop a need to visit an emergency department (ED). Such visits are unanticipated and unwelcome events and constitute utilization of an expensive component of the healthcare system. ED visits during OPAT have not received as much attention as hospitalizations. In one study that examined ED visits for patients on OPAT at a tertiary care centre in Canada, 43% of 104 patients discharged home on OPAT visited an ED within 60 days.7 An excess of ED visits stresses the healthcare system. Elucidation of the frequency of ED visits during OPAT will help quantify the magnitude of this stress. Identification of risk factors that result in ED visits may help identify opportunities for preventing them. No prior study has comprehensively examined demographic and clinical characteristics of OPAT and their association with ED visits. The purpose of this study was to describe ED visits during OPAT, to identify risk factors for ED visits and to identify risk factors for hospitalization among patients who visit the ED. Patients and methods Study design and setting The study was a retrospective cohort study conducted at Cleveland Clinic, a large multispecialty tertiary care referral centre in the USA. The Cleveland Clinic Health System (CCHS) includes the Cleveland Clinic Main Campus (a 1400 bed hospital), 10 regional hospitals in north-east Ohio and over 150 outpatient locations in northern Ohio, including 18 full-service family health centres and 21 express and urgent care sites. The entire health system uses a common electronic medical record (EMR), allowing for examination of patient care data throughout the system. Cleveland Clinic Main Campus has a large OPAT programme that has been in existence since 1979 and now encompasses ∼2500 OPAT courses a year. Initiation of OPAT at Cleveland Clinic requires evaluation by a Cleveland Clinic infectious disease attending physician.8 An electronic structured start-of-care form is completed by the infectious disease physician at the start of the OPAT course. The Cleveland Clinic OPAT registry identifies patients through the OPAT start-of-care forms. The common EMR allows for identification of ED visits not just to Cleveland Clinic Main Campus but to the entire CCHS. Screening, inclusion and exclusion criteria All OPAT courses for patients aged >18 years started between 1 January 2013 and 1 January 2017 at Cleveland Clinic Main Campus were identified from the Cleveland Clinic OPAT registry and screened for inclusion in this study. The first OPAT course per patient during the study period was included. OPAT courses that lasted <1 day were excluded. Data sources and data extraction Sources of data were the Cleveland Clinic OPAT registry and the Cleveland Clinic enterprise data vault (EDV). The Cleveland Clinic OPAT registry was used to identify OPAT episodes within the designated time frame, as well as the site of infection, antimicrobials, vascular access device, OPAT initiation date, the OPAT delivery site and planned duration of therapy. The comorbid conditions, residence ZIP code, county, health insurance and prior ED visits were obtained from the EDV. For the included patients, ED visits within the CCHS that occurred after the OPAT start date were searched for in the EDV and those that occurred within 30 days were identified. The reason for ED visit and patient disposition after the ED visit were ascertained by manual review of the EMR. For the sake of consistency, this was done by a single investigator. Study data were collected in a relational database (Microsoft® Access). Ethics The study was approved by the Cleveland Clinic institutional review board (IRB no. 16-1692). A waiver of the requirement for patient consent was granted. Definitions Age was defined as patient age on the OPAT start date. Comorbid conditions included diabetes mellitus, end-stage renal disease (on haemodialysis), COPD, liver cirrhosis and malignancy. Malignancy was defined as a history of haematological malignancy or solid tumour. The OPAT initiation site was ‘inpatient’ if the OPAT was begun during hospitalization and ‘outpatient’ if it was started in the outpatient setting. The OPAT delivery site could be home, residential facility or non-residential facility. Skilled nursing facilities and long-term acute care facilities were considered residential facilities. Infusion centres and dialysis centres were considered non-residential facilities. The county of residence was grouped into one of the following: Cuyahoga (county where Cleveland Clinic Main Campus is located), surrounding (county that has a border with Cuyahoga County), distant (a US county that does not border Cuyahoga County) and foreign (outside the USA). Medicare and Medicaid were considered government insurance. The mean annual household income for the year 2015 for the ZIP code in which the patient resided was taken to be the estimated patient income; it was calculated from publicly available data obtained from the Internal Revenue Service website.9 ED visits resulting from the infectious disease being treated, antimicrobial adverse events or vascular access problems were categorized as OPAT-related ED visits. Outcomes The primary outcome was days to ED visit. Death and hospital admission were recognized as competing outcomes that would preclude an ED visit. An ED visit that led to hospital admission was counted as an ED visit. Statistical analysis Baseline characteristics of OPAT courses with or without at least one ED visit were compared. Continuous variables were compared using Student’s t-test and categorical variables using the χ2 test. A frequency distribution of the reason for ED visit was tabulated. Risk factors for ED visits were identified in a multivariable Cox proportional hazards regression model by examining the association of relevant demographic and clinical factors, which were felt to possibly influence need for an ED visit, with the first ED visit per patient. The outcome examined was time to ED visit. The time-to-event variable was censored at 30 days. For patients who died or were subsequently hospitalized within 30 days, before they had a single ED visit, the variable was censored on the date of death or the date of hospitalization, respectively. Explanatory variables included in the initial model were patient age (decade of age), gender, calendar year, OPAT initiation site, OPAT delivery site, county of residence, insurance (government, private, international, uninsured), income quartile, diagnoses being treated with OPAT, parenteral antimicrobials administered, log anticipated OPAT duration and prior ED visit within the previous year. Variable selection was then done by backward stepwise elimination of the variable that caused the greatest reduction in the Akaike Information Criterion (AIC) at each step until the optimal model was obtained, using the stepAIC function in the R package MASS.10 Outliers were investigated for errors by examining case deletion residuals. The distribution of martingale residuals across values or levels of variables included in the final model was checked and the residuals were found to be evenly distributed. The proportional hazards assumption was checked using Schoenfeld residuals and the model was not in violation. To account for the competing risks of subsequent hospitalization or death, a subdistribution proportional hazards competing risks regression model was developed, with the variables in the final Cox proportional hazards model as the covariates, using the method of Fine and Gray.11,12 For the first ED visit per patient, for patients who did not have an intervening hospitalization, factors associated with hospital admission were examined in a multivariable logistic regression model including all the baseline variables as covariates. Variable selection was then done by backward stepwise selection to identify the optimal model based on identifying the model with the highest R2 for each given number of predictors and the model with the lowest AIC among these models.10 The final model was checked for multicollinearity using generalized variance inflation factors13,14 and there was no evidence of it. The final model was checked for interaction effects and none was found. Statistical analyses were done using R15 using the core functions in the base package and the packages tableone,16 survival,17 MASS,10 cmprsk12 and car.14 Results Between 1 January 2013 and 1 January 2017, 8263 patients were treated with a total of 11 391 OPAT courses. Of the 8263 OPAT courses included in the study (first OPAT course per patient), 416 (5%) had at least one ED visit within 30 days of initiation of OPAT, the first ED visit occurring a median of 10 days from the OPAT start date. Thirty-five had an intervening hospitalization between the OPAT start date and the first subsequent ED visit. Thus, 381 patients had at least one ED visit without an intervening hospitalization. Baseline characteristics of included patients The mean age at OPAT was 59 years (SD 16 years), 4737 (57%) were male and the median anticipated duration of OPAT was 20 days (IQR 11–34 days). The characteristics of the included patients are outlined in Table 1. Table 1. Baseline characteristics of included OPAT courses Characteristica ED visitb, N = 381 No ED visit, N = 7882 Pc Patient age (years), mean (SD) 57 (17) 59 (16) 0.004 Male, n (%) 189 (50) 4548 (58) 0.002 Year, n (%) <0.001  2013 215 (56) 2161 (27)  2014 76 (20) 1952 (25)  2015 45 (12) 1910 (24)  2016 45 (12) 1859 (24) Comorbid conditions, n (%)  diabetes mellitus 138 (36) 2817 (36) 0.891  end-stage renal disease 52 (14) 801 (10) 0.036  cirrhosis of liver 26 (7) 423 (5) 0.267  COPD 105 (28) 1615 (21) 0.001  malignancy 263 (69) 3964 (50) <0.001 Infection being treated, n (%)  abdominal 49 (13) 1014 (13) 1.000  cardiovascular 69 (18) 1544 (20) 0.519  CNS 23 (6) 538 (7) 0.622  genitourinary 34 (9) 491 (6) 0.046  head and neck 6 (2) 132 (2) 1.000  osteoarticular 104 (27) 2193 (28) 0.869  skin and skin structure 47 (12) 871 (11) 0.486  thoracic 23 (6) 625 (8) 0.213  vascular catheter infection/ unspecified bacteraemia 44 (12) 1014 (13) 0.501  other 6 (2) 89 (1) 0.582 Parenteral antimicrobial, n (%)  penicillins 127 (33) 2235 (28) 0.041  cephalosporins 66 (17) 1493 (19) 0.470  carbapenems 52 (14) 1232 (16) 0.332  glycopeptides 141 (37) 2822 (36) 0.671  lipopeptides 23 (6) 462 (6) 0.976  other antibacterials 23 (6) 395 (5) 0.440  antifungals 19 (5) 393 (5) 0.912  antivirals 7 (2) 201 (3) 0.484 Outpatient OPAT initiation, n (%) 18 (5) 419 (5) 0.699 Log anticipated OPAT duration (days), mean (SD) 2.79 (0.79) 2.88 (0.76) 0.024 OPAT delivery site, n (%) 0.152  home 220 (58) 4517 (57)  residential facility 142 (37) 3033 (39)  dialysis centre 19 (5) 274 (4)  infusion centre 0 (0) 58 (0.7) County of residence, n (%) <0.001  Cuyahoga 196 (51) 2407 (31)  surrounding 105 (28) 1517 (19)  distant 77 (20) 3890 (49)  foreign 3 (0.8) 68 (0.9) Insurance, n (%) <0.001  government 274 (72) 4919 (62)  private 79 (21) 2522 (32)  international 2 (0.5) 51 (0.6)  uninsured 26 (7) 390 (5) Income quartile (annual income), n (%) <0.001  USD 19 300–41 400 120 (32) 1838 (23)  USD 41 400–51 500 56 (15) 1919 (24)  USD 51 500–65 400 86 (23) 1889 (24)  USD 65 400–588 000 99 (26) 1915 (24)  unknown 20 (5) 321 (4) Vascular access device, n (%) 0.563  PICC 296 (78) 6025 (76)  Hohn 28 (7) 698 (9)  Hickman 16 (4) 336 (4)  indwelling port 11 (3) 296 (4)  midline 5 (1) 105 (1)  tunnelled dialysis catheter 21 (6) 303 (4)  other 4 (1) 119 (2) Infectious disease staff affiliationd, n (%) 0.435  bone and joint 111 (29) 2154 (27)  transplant 102 (27) 2106 (27)  neurological 59 (16) 1314 (17)  cardiac 47 (12) 1193 (15)  other 62 (16) 1115 (14) Prior ED visit within 1 year, n (%) 292 (77) 3640 (46) <0.001 Characteristica ED visitb, N = 381 No ED visit, N = 7882 Pc Patient age (years), mean (SD) 57 (17) 59 (16) 0.004 Male, n (%) 189 (50) 4548 (58) 0.002 Year, n (%) <0.001  2013 215 (56) 2161 (27)  2014 76 (20) 1952 (25)  2015 45 (12) 1910 (24)  2016 45 (12) 1859 (24) Comorbid conditions, n (%)  diabetes mellitus 138 (36) 2817 (36) 0.891  end-stage renal disease 52 (14) 801 (10) 0.036  cirrhosis of liver 26 (7) 423 (5) 0.267  COPD 105 (28) 1615 (21) 0.001  malignancy 263 (69) 3964 (50) <0.001 Infection being treated, n (%)  abdominal 49 (13) 1014 (13) 1.000  cardiovascular 69 (18) 1544 (20) 0.519  CNS 23 (6) 538 (7) 0.622  genitourinary 34 (9) 491 (6) 0.046  head and neck 6 (2) 132 (2) 1.000  osteoarticular 104 (27) 2193 (28) 0.869  skin and skin structure 47 (12) 871 (11) 0.486  thoracic 23 (6) 625 (8) 0.213  vascular catheter infection/ unspecified bacteraemia 44 (12) 1014 (13) 0.501  other 6 (2) 89 (1) 0.582 Parenteral antimicrobial, n (%)  penicillins 127 (33) 2235 (28) 0.041  cephalosporins 66 (17) 1493 (19) 0.470  carbapenems 52 (14) 1232 (16) 0.332  glycopeptides 141 (37) 2822 (36) 0.671  lipopeptides 23 (6) 462 (6) 0.976  other antibacterials 23 (6) 395 (5) 0.440  antifungals 19 (5) 393 (5) 0.912  antivirals 7 (2) 201 (3) 0.484 Outpatient OPAT initiation, n (%) 18 (5) 419 (5) 0.699 Log anticipated OPAT duration (days), mean (SD) 2.79 (0.79) 2.88 (0.76) 0.024 OPAT delivery site, n (%) 0.152  home 220 (58) 4517 (57)  residential facility 142 (37) 3033 (39)  dialysis centre 19 (5) 274 (4)  infusion centre 0 (0) 58 (0.7) County of residence, n (%) <0.001  Cuyahoga 196 (51) 2407 (31)  surrounding 105 (28) 1517 (19)  distant 77 (20) 3890 (49)  foreign 3 (0.8) 68 (0.9) Insurance, n (%) <0.001  government 274 (72) 4919 (62)  private 79 (21) 2522 (32)  international 2 (0.5) 51 (0.6)  uninsured 26 (7) 390 (5) Income quartile (annual income), n (%) <0.001  USD 19 300–41 400 120 (32) 1838 (23)  USD 41 400–51 500 56 (15) 1919 (24)  USD 51 500–65 400 86 (23) 1889 (24)  USD 65 400–588 000 99 (26) 1915 (24)  unknown 20 (5) 321 (4) Vascular access device, n (%) 0.563  PICC 296 (78) 6025 (76)  Hohn 28 (7) 698 (9)  Hickman 16 (4) 336 (4)  indwelling port 11 (3) 296 (4)  midline 5 (1) 105 (1)  tunnelled dialysis catheter 21 (6) 303 (4)  other 4 (1) 119 (2) Infectious disease staff affiliationd, n (%) 0.435  bone and joint 111 (29) 2154 (27)  transplant 102 (27) 2106 (27)  neurological 59 (16) 1314 (17)  cardiac 47 (12) 1193 (15)  other 62 (16) 1115 (14) Prior ED visit within 1 year, n (%) 292 (77) 3640 (46) <0.001 PICC, percutaneously inserted central catheter. a Expressed as mean (SD) for continuous variables and n (%) for categorical variables. b At least one ED visit within 30 days of initiation of OPAT. c Student’s t-test for continuous variables and the χ2 test for categorical variables. d Subspecialty affiliation of the infectious disease staff physician responsible for the OPAT course. Table 1. Baseline characteristics of included OPAT courses Characteristica ED visitb, N = 381 No ED visit, N = 7882 Pc Patient age (years), mean (SD) 57 (17) 59 (16) 0.004 Male, n (%) 189 (50) 4548 (58) 0.002 Year, n (%) <0.001  2013 215 (56) 2161 (27)  2014 76 (20) 1952 (25)  2015 45 (12) 1910 (24)  2016 45 (12) 1859 (24) Comorbid conditions, n (%)  diabetes mellitus 138 (36) 2817 (36) 0.891  end-stage renal disease 52 (14) 801 (10) 0.036  cirrhosis of liver 26 (7) 423 (5) 0.267  COPD 105 (28) 1615 (21) 0.001  malignancy 263 (69) 3964 (50) <0.001 Infection being treated, n (%)  abdominal 49 (13) 1014 (13) 1.000  cardiovascular 69 (18) 1544 (20) 0.519  CNS 23 (6) 538 (7) 0.622  genitourinary 34 (9) 491 (6) 0.046  head and neck 6 (2) 132 (2) 1.000  osteoarticular 104 (27) 2193 (28) 0.869  skin and skin structure 47 (12) 871 (11) 0.486  thoracic 23 (6) 625 (8) 0.213  vascular catheter infection/ unspecified bacteraemia 44 (12) 1014 (13) 0.501  other 6 (2) 89 (1) 0.582 Parenteral antimicrobial, n (%)  penicillins 127 (33) 2235 (28) 0.041  cephalosporins 66 (17) 1493 (19) 0.470  carbapenems 52 (14) 1232 (16) 0.332  glycopeptides 141 (37) 2822 (36) 0.671  lipopeptides 23 (6) 462 (6) 0.976  other antibacterials 23 (6) 395 (5) 0.440  antifungals 19 (5) 393 (5) 0.912  antivirals 7 (2) 201 (3) 0.484 Outpatient OPAT initiation, n (%) 18 (5) 419 (5) 0.699 Log anticipated OPAT duration (days), mean (SD) 2.79 (0.79) 2.88 (0.76) 0.024 OPAT delivery site, n (%) 0.152  home 220 (58) 4517 (57)  residential facility 142 (37) 3033 (39)  dialysis centre 19 (5) 274 (4)  infusion centre 0 (0) 58 (0.7) County of residence, n (%) <0.001  Cuyahoga 196 (51) 2407 (31)  surrounding 105 (28) 1517 (19)  distant 77 (20) 3890 (49)  foreign 3 (0.8) 68 (0.9) Insurance, n (%) <0.001  government 274 (72) 4919 (62)  private 79 (21) 2522 (32)  international 2 (0.5) 51 (0.6)  uninsured 26 (7) 390 (5) Income quartile (annual income), n (%) <0.001  USD 19 300–41 400 120 (32) 1838 (23)  USD 41 400–51 500 56 (15) 1919 (24)  USD 51 500–65 400 86 (23) 1889 (24)  USD 65 400–588 000 99 (26) 1915 (24)  unknown 20 (5) 321 (4) Vascular access device, n (%) 0.563  PICC 296 (78) 6025 (76)  Hohn 28 (7) 698 (9)  Hickman 16 (4) 336 (4)  indwelling port 11 (3) 296 (4)  midline 5 (1) 105 (1)  tunnelled dialysis catheter 21 (6) 303 (4)  other 4 (1) 119 (2) Infectious disease staff affiliationd, n (%) 0.435  bone and joint 111 (29) 2154 (27)  transplant 102 (27) 2106 (27)  neurological 59 (16) 1314 (17)  cardiac 47 (12) 1193 (15)  other 62 (16) 1115 (14) Prior ED visit within 1 year, n (%) 292 (77) 3640 (46) <0.001 Characteristica ED visitb, N = 381 No ED visit, N = 7882 Pc Patient age (years), mean (SD) 57 (17) 59 (16) 0.004 Male, n (%) 189 (50) 4548 (58) 0.002 Year, n (%) <0.001  2013 215 (56) 2161 (27)  2014 76 (20) 1952 (25)  2015 45 (12) 1910 (24)  2016 45 (12) 1859 (24) Comorbid conditions, n (%)  diabetes mellitus 138 (36) 2817 (36) 0.891  end-stage renal disease 52 (14) 801 (10) 0.036  cirrhosis of liver 26 (7) 423 (5) 0.267  COPD 105 (28) 1615 (21) 0.001  malignancy 263 (69) 3964 (50) <0.001 Infection being treated, n (%)  abdominal 49 (13) 1014 (13) 1.000  cardiovascular 69 (18) 1544 (20) 0.519  CNS 23 (6) 538 (7) 0.622  genitourinary 34 (9) 491 (6) 0.046  head and neck 6 (2) 132 (2) 1.000  osteoarticular 104 (27) 2193 (28) 0.869  skin and skin structure 47 (12) 871 (11) 0.486  thoracic 23 (6) 625 (8) 0.213  vascular catheter infection/ unspecified bacteraemia 44 (12) 1014 (13) 0.501  other 6 (2) 89 (1) 0.582 Parenteral antimicrobial, n (%)  penicillins 127 (33) 2235 (28) 0.041  cephalosporins 66 (17) 1493 (19) 0.470  carbapenems 52 (14) 1232 (16) 0.332  glycopeptides 141 (37) 2822 (36) 0.671  lipopeptides 23 (6) 462 (6) 0.976  other antibacterials 23 (6) 395 (5) 0.440  antifungals 19 (5) 393 (5) 0.912  antivirals 7 (2) 201 (3) 0.484 Outpatient OPAT initiation, n (%) 18 (5) 419 (5) 0.699 Log anticipated OPAT duration (days), mean (SD) 2.79 (0.79) 2.88 (0.76) 0.024 OPAT delivery site, n (%) 0.152  home 220 (58) 4517 (57)  residential facility 142 (37) 3033 (39)  dialysis centre 19 (5) 274 (4)  infusion centre 0 (0) 58 (0.7) County of residence, n (%) <0.001  Cuyahoga 196 (51) 2407 (31)  surrounding 105 (28) 1517 (19)  distant 77 (20) 3890 (49)  foreign 3 (0.8) 68 (0.9) Insurance, n (%) <0.001  government 274 (72) 4919 (62)  private 79 (21) 2522 (32)  international 2 (0.5) 51 (0.6)  uninsured 26 (7) 390 (5) Income quartile (annual income), n (%) <0.001  USD 19 300–41 400 120 (32) 1838 (23)  USD 41 400–51 500 56 (15) 1919 (24)  USD 51 500–65 400 86 (23) 1889 (24)  USD 65 400–588 000 99 (26) 1915 (24)  unknown 20 (5) 321 (4) Vascular access device, n (%) 0.563  PICC 296 (78) 6025 (76)  Hohn 28 (7) 698 (9)  Hickman 16 (4) 336 (4)  indwelling port 11 (3) 296 (4)  midline 5 (1) 105 (1)  tunnelled dialysis catheter 21 (6) 303 (4)  other 4 (1) 119 (2) Infectious disease staff affiliationd, n (%) 0.435  bone and joint 111 (29) 2154 (27)  transplant 102 (27) 2106 (27)  neurological 59 (16) 1314 (17)  cardiac 47 (12) 1193 (15)  other 62 (16) 1115 (14) Prior ED visit within 1 year, n (%) 292 (77) 3640 (46) <0.001 PICC, percutaneously inserted central catheter. a Expressed as mean (SD) for continuous variables and n (%) for categorical variables. b At least one ED visit within 30 days of initiation of OPAT. c Student’s t-test for continuous variables and the χ2 test for categorical variables. d Subspecialty affiliation of the infectious disease staff physician responsible for the OPAT course. Deaths, hospital admissions and ED visits Three hundred and eighty-one patients (4.6%) had at least one ED visit within 30 days of OPAT initiation in the absence of an intervening hospitalization. In addition, 50 patients (0.6%) died and 1133 (14%) were hospitalized within 30 days without an intervening ED visit. The latter were patients who were admitted to hospital directly, not through the ED. The cumulative probabilities of having these competing events (ED visit, hospital admission and death) over time are shown in Figure 1. Figure 1. View largeDownload slide Cumulative occurrence of competing events (ED visit, hospital admission and death) in the 30 days following initiation of OPAT. Figure 1. View largeDownload slide Cumulative occurrence of competing events (ED visit, hospital admission and death) in the 30 days following initiation of OPAT. Description of ED visits Of the 381 ED visits (first ED visit for patients who had at least one ED visit without an intervening hospitalization), 193 (51%) were OPAT related. The most common visit reason was vascular access complication, which accounted for 104 ED visits (54% of OPAT-related ED visits, 27% of all ED visits). The most common vascular access complications were occlusion and dislodgement, which accounted for 53 and 19 ED visits, respectively. The proportion of OPAT courses with at least one ED visit were 9.1%, 3.8%, 2.3% and 2.4%, respectively, for the years 2013, 2014, 2015 and 2016. Reasons for ED visit are outlined in Table 2. Table 2. Frequency distribution of reason for ED visit ED visit reason n Percentage of ED visits OPAT related  worsening infectiona 18 5  antimicrobial adverse event 14 4  vascular access complication 104 27  symptom associated with OPAT diagnosisb 57 15 Non-OPAT related  exacerbation of a chronic condition 43 11  complication of a chronic condition 24 6  new non-OPAT problem 121 32 ED visit reason n Percentage of ED visits OPAT related  worsening infectiona 18 5  antimicrobial adverse event 14 4  vascular access complication 104 27  symptom associated with OPAT diagnosisb 57 15 Non-OPAT related  exacerbation of a chronic condition 43 11  complication of a chronic condition 24 6  new non-OPAT problem 121 32 a Symptom(s) deemed to be from worsening infection after clinical evaluation. b Symptoms related to the infection being treated but deemed not to be worsening infection after clinical evaluation. Table 2. Frequency distribution of reason for ED visit ED visit reason n Percentage of ED visits OPAT related  worsening infectiona 18 5  antimicrobial adverse event 14 4  vascular access complication 104 27  symptom associated with OPAT diagnosisb 57 15 Non-OPAT related  exacerbation of a chronic condition 43 11  complication of a chronic condition 24 6  new non-OPAT problem 121 32 ED visit reason n Percentage of ED visits OPAT related  worsening infectiona 18 5  antimicrobial adverse event 14 4  vascular access complication 104 27  symptom associated with OPAT diagnosisb 57 15 Non-OPAT related  exacerbation of a chronic condition 43 11  complication of a chronic condition 24 6  new non-OPAT problem 121 32 a Symptom(s) deemed to be from worsening infection after clinical evaluation. b Symptoms related to the infection being treated but deemed not to be worsening infection after clinical evaluation. Risk factors for ED visits Death and hospital admission were recognized as competing outcomes for ED visit and this was accounted for in the analysis. In a multivariable subdistribution proportional hazards competing risks regression model, the variables that were significantly associated with having an ED visit are shown in Table 3. Notably, patients with a prior ED visit within the preceding year had a 2.3-fold higher hazard of having an ED visit within 30 days of starting OPAT than patients who had not been to an ED in the preceding year (HR 2.29, 95% CI 1.76–2.98, P = 8.1 × 10−10). Other factors associated with a significantly higher hazard of ED visits were female sex (HR 1.25, 95% CI 1.02–1.53, P = 0.032), malignancy (HR 1.49, 95% CI 1.19–1.87, P = 6.1 × 10−4) and treatment with a parenteral penicillin (HR 1.31, 95% CI 1.06–1.61, P = 0.014). Factors associated with a significantly lower hazard of ED visits were higher age (HR 0.89 per decade of age, 95% CI 0.83–0.95, P = 4.0 × 10−4), distant county as opposed to Cuyahoga County (HR 0.42, 95% CI 0.31–0.56, P = 8.5 × 10−9) and private insurance as opposed to government insurance (HR 0.59, 95% CI 0.46–0.77, P = 6.3 × 10−5). The hazard for ED visit was lower for each subsequent year compared with the year 2013. Table 3. Factors significantly associated with ED visits, for patients on OPAT, in a multivariable subdistribution proportional hazards competing risks regression model Factora HR 95% CI P Age (decade) 0.89 0.83–0.95 4.0 × 10−4 Female 1.25 1.02–1.53 0.032 Malignancy 1.49 1.19–1.87 6.1 × 10−4 Treatment with a parenteral penicillin 1.31 1.06–1.61 0.014 County (reference: Cuyahoga)  surrounding 1.04 0.81–1.32 0.78  distant 0.42 0.31–0.56 8.5 × 10−9  foreign 0.87 0.22–3.39 0.84 Insurance (reference: government)  private 0.59 0.46–0.77 6.3 × 10−5  international 0.79 0.16–3.92 0.77  uninsured 1.16 0.75–1.80 0.50 Year (reference: 2013)  2014 0.42 0.32–0.54 6.8 × 10−11  2015 0.25 0.18–0.35 <2.2 × 10−16  2016 0.27 0.20–0.37 1.6 × 10−15 Prior ED visitb 2.29 1.76–2.98 8.1 × 10−10 Factora HR 95% CI P Age (decade) 0.89 0.83–0.95 4.0 × 10−4 Female 1.25 1.02–1.53 0.032 Malignancy 1.49 1.19–1.87 6.1 × 10−4 Treatment with a parenteral penicillin 1.31 1.06–1.61 0.014 County (reference: Cuyahoga)  surrounding 1.04 0.81–1.32 0.78  distant 0.42 0.31–0.56 8.5 × 10−9  foreign 0.87 0.22–3.39 0.84 Insurance (reference: government)  private 0.59 0.46–0.77 6.3 × 10−5  international 0.79 0.16–3.92 0.77  uninsured 1.16 0.75–1.80 0.50 Year (reference: 2013)  2014 0.42 0.32–0.54 6.8 × 10−11  2015 0.25 0.18–0.35 <2.2 × 10−16  2016 0.27 0.20–0.37 1.6 × 10−15 Prior ED visitb 2.29 1.76–2.98 8.1 × 10−10 a COPD (P = 0.27), outpatient initiation (P = 0.27) and log anticipated OPAT duration (P = 0.45) were also included as variables in the final model but were not statistically significant. b Prior ED visit within the preceding year. Table 3. Factors significantly associated with ED visits, for patients on OPAT, in a multivariable subdistribution proportional hazards competing risks regression model Factora HR 95% CI P Age (decade) 0.89 0.83–0.95 4.0 × 10−4 Female 1.25 1.02–1.53 0.032 Malignancy 1.49 1.19–1.87 6.1 × 10−4 Treatment with a parenteral penicillin 1.31 1.06–1.61 0.014 County (reference: Cuyahoga)  surrounding 1.04 0.81–1.32 0.78  distant 0.42 0.31–0.56 8.5 × 10−9  foreign 0.87 0.22–3.39 0.84 Insurance (reference: government)  private 0.59 0.46–0.77 6.3 × 10−5  international 0.79 0.16–3.92 0.77  uninsured 1.16 0.75–1.80 0.50 Year (reference: 2013)  2014 0.42 0.32–0.54 6.8 × 10−11  2015 0.25 0.18–0.35 <2.2 × 10−16  2016 0.27 0.20–0.37 1.6 × 10−15 Prior ED visitb 2.29 1.76–2.98 8.1 × 10−10 Factora HR 95% CI P Age (decade) 0.89 0.83–0.95 4.0 × 10−4 Female 1.25 1.02–1.53 0.032 Malignancy 1.49 1.19–1.87 6.1 × 10−4 Treatment with a parenteral penicillin 1.31 1.06–1.61 0.014 County (reference: Cuyahoga)  surrounding 1.04 0.81–1.32 0.78  distant 0.42 0.31–0.56 8.5 × 10−9  foreign 0.87 0.22–3.39 0.84 Insurance (reference: government)  private 0.59 0.46–0.77 6.3 × 10−5  international 0.79 0.16–3.92 0.77  uninsured 1.16 0.75–1.80 0.50 Year (reference: 2013)  2014 0.42 0.32–0.54 6.8 × 10−11  2015 0.25 0.18–0.35 <2.2 × 10−16  2016 0.27 0.20–0.37 1.6 × 10−15 Prior ED visitb 2.29 1.76–2.98 8.1 × 10−10 a COPD (P = 0.27), outpatient initiation (P = 0.27) and log anticipated OPAT duration (P = 0.45) were also included as variables in the final model but were not statistically significant. b Prior ED visit within the preceding year. Factors associated with hospital admission for patients who had ED visits Among the 381 ED visits, 85 (22%) resulted in hospital admission. The factors that were independently associated with hospital admission are outlined in Table 4. Among patients who visited the ED within 30 days of starting OPAT, patients on treatment with a parenteral antiviral agent had 33-fold higher odds of being admitted than patients not on parenteral antiviral agents (OR 33.37, 95% CI 3.98–384.71, P = 0.002). Compared with patients who visited the ED for non-OPAT-related reasons, those presenting with worsening infection had 19-fold higher odds (OR 18.95, 95% CI 5.50–79.85, P = 1.2 × 10−7) and those presenting with vascular access problems had 3-fold lower odds (OR 0.36, 95% CI 0.14–0.83, P = 0.022) of being admitted to hospital. The decision to admit to hospital from the ED was also influenced significantly by which ED the patient presented to. Table 4. Significant associations with hospital admission, for ED visits, in a multivariable logistic regression model Factora OR 95% CI P Log anticipated OPAT duration 0.66 0.45–0.96 0.031 Parenteral antiviral treatment 33.37 3.98–384.71 0.002 ED (reference: Cleveland Clinic Main Campus)b  Ashtabula (NT and 80 kmc) 3.64 1.06–11.64 0.033  Avon (NT and 30 kmc) 9.03 2.96–28.02 1.1 × 10−4  Fairview (T and 18 kmc) 3.35 1.12–9.44 0.024  Marymount (NT and 9 kmc) 3.82 1.01–12.96 0.037  Medina (NT and 44 kmc) 9.63 3.96–24.29 8.6 × 10−7  Twinsburg (NT and 27 kmc) 9.32 2.79–31.27 2.6 × 10−4 ED visit reason (reference: non-OPAT reason)d  vascular access problem 0.36 0.14–0.83 0.022  worsening infection 18.95 5.50–79.85 1.2 × 10−7 Factora OR 95% CI P Log anticipated OPAT duration 0.66 0.45–0.96 0.031 Parenteral antiviral treatment 33.37 3.98–384.71 0.002 ED (reference: Cleveland Clinic Main Campus)b  Ashtabula (NT and 80 kmc) 3.64 1.06–11.64 0.033  Avon (NT and 30 kmc) 9.03 2.96–28.02 1.1 × 10−4  Fairview (T and 18 kmc) 3.35 1.12–9.44 0.024  Marymount (NT and 9 kmc) 3.82 1.01–12.96 0.037  Medina (NT and 44 kmc) 9.63 3.96–24.29 8.6 × 10−7  Twinsburg (NT and 27 kmc) 9.32 2.79–31.27 2.6 × 10−4 ED visit reason (reference: non-OPAT reason)d  vascular access problem 0.36 0.14–0.83 0.022  worsening infection 18.95 5.50–79.85 1.2 × 10−7 a Age (P = 0.06), diabetes mellitus (P = 0.10) and OPAT site (P = 0.06) were also included as variables in the final model but were not statistically significant. b Lakewood (P = 0.38), Lutheran (P = 0.29) and other EDs (P = 0.29) not significant. c Institution teaching status (T = teaching and NT = non-teaching) and distance from Cleveland Clinic Main Campus. d Antimicrobial adverse event (P = 0.49) and symptoms associated with OPAT diagnosis (P = 0.16) not significant. Table 4. Significant associations with hospital admission, for ED visits, in a multivariable logistic regression model Factora OR 95% CI P Log anticipated OPAT duration 0.66 0.45–0.96 0.031 Parenteral antiviral treatment 33.37 3.98–384.71 0.002 ED (reference: Cleveland Clinic Main Campus)b  Ashtabula (NT and 80 kmc) 3.64 1.06–11.64 0.033  Avon (NT and 30 kmc) 9.03 2.96–28.02 1.1 × 10−4  Fairview (T and 18 kmc) 3.35 1.12–9.44 0.024  Marymount (NT and 9 kmc) 3.82 1.01–12.96 0.037  Medina (NT and 44 kmc) 9.63 3.96–24.29 8.6 × 10−7  Twinsburg (NT and 27 kmc) 9.32 2.79–31.27 2.6 × 10−4 ED visit reason (reference: non-OPAT reason)d  vascular access problem 0.36 0.14–0.83 0.022  worsening infection 18.95 5.50–79.85 1.2 × 10−7 Factora OR 95% CI P Log anticipated OPAT duration 0.66 0.45–0.96 0.031 Parenteral antiviral treatment 33.37 3.98–384.71 0.002 ED (reference: Cleveland Clinic Main Campus)b  Ashtabula (NT and 80 kmc) 3.64 1.06–11.64 0.033  Avon (NT and 30 kmc) 9.03 2.96–28.02 1.1 × 10−4  Fairview (T and 18 kmc) 3.35 1.12–9.44 0.024  Marymount (NT and 9 kmc) 3.82 1.01–12.96 0.037  Medina (NT and 44 kmc) 9.63 3.96–24.29 8.6 × 10−7  Twinsburg (NT and 27 kmc) 9.32 2.79–31.27 2.6 × 10−4 ED visit reason (reference: non-OPAT reason)d  vascular access problem 0.36 0.14–0.83 0.022  worsening infection 18.95 5.50–79.85 1.2 × 10−7 a Age (P = 0.06), diabetes mellitus (P = 0.10) and OPAT site (P = 0.06) were also included as variables in the final model but were not statistically significant. b Lakewood (P = 0.38), Lutheran (P = 0.29) and other EDs (P = 0.29) not significant. c Institution teaching status (T = teaching and NT = non-teaching) and distance from Cleveland Clinic Main Campus. d Antimicrobial adverse event (P = 0.49) and symptoms associated with OPAT diagnosis (P = 0.16) not significant. Discussion This study finds that, excluding patients who were admitted directly to hospital without passing through the ED, about 5% of patients on OPAT have at least one ED visit within the 30 days following initiation of OPAT. Patients with an ED visit within the previous year have a 2.3-fold higher hazard of having an ED visit than patients without an ED visit in the preceding year. This is probably partly a marker of access to care. Some patients have no other recourse than to visit an ED when faced with a medical problem that requires attention. It is also possible that some patients with repeated ED visits are those who have medical conditions that lead to more medical emergencies. About half of the ED visits are for non-OPAT-related reasons. This is a reflection of the underlying comorbidity burden in many patients who develop a condition that requires parenteral antimicrobial therapy. The study also shows that the decision to admit to hospital from the ED varies significantly by the reason for the ED visit. Patients visiting the ED for vascular access problems are much less likely to be admitted to hospital than patients visiting for worsening infection or non-OPAT-related reasons. This is probably because vascular access problems that bring patients to the ED can usually be addressed without having to admit the patient to the hospital. The proportion of patients with an ED visit in our study (5% within 30 days) is much lower than was noted in a former study in Canada (43% within 60 days).7 This difference can be explained by the difference in the study setting and the fact that duration of follow-up in the Canadian study was twice as long. The Canadian study was done in a centre without a formal OPAT programme and only half of the patients had an infectious disease referral. Our study was done in an institution with a well-organized OPAT programme where every patient is seen by an infectious disease physician. A notable finding in our study is a drop in the proportion of patients with ED visits from 9% in 2013 to ∼2.4% in 2015 and 2016. One of the factors that might have contributed to this decline was an initiative taken by the most frequently used home healthcare provider to manage catheter occlusions by administering recombinant tissue plasminogen activator in patients’ homes instead of sending them to the ED, a practice that was also soon adopted by other agencies. But this alone would not be an adequate explanation for this reduction. It is very likely that concomitant efforts by the institution to reduce hospital readmissions (a response to the introduction of financial penalties for 30 day readmissions by government payors) also inadvertently resulted in reduction of ED visits. The biggest strength of this study is the large sample size. It allowed for use of more comprehensive models including many relevant variables that would not have been possible with a smaller study sample. Our OPAT programme and OPAT registry allowed for an accurate identification of all OPAT courses within our hospital.8 Including all ED visits across the health system allowed for a more comprehensive picture of the ED visit burden than would have been possible if we had only examined ED visits in our own hospital. In a rapidly changing healthcare environment, incentives to visit or avoid the ED change with time and could influence the results of any study examining ED visits. To control for this factor we included calendar year as a covariate in our models, thereby adjusting for policy or process factors that might have changed over time. The study has its limitations. We were only able to identify ED visits within the CCHS. There were certainly ED visits that occurred outside the CCHS, particularly for patients who lived at great distances from Cleveland Clinic. Cleveland Clinic Hospital is located in Cuyahoga County and all the EDs of the CCHS are located in Cuyahoga County or the surrounding counties. To control for the likelihood of visiting one of the EDs within the CCHS, we adjusted for the patients’ county of residence in our models to account for the fact that patients who lived closer to the hospitals of the CCHS were more likely to visit one of the EDs associated with the CCHS. We also recognized that an important reason for patients to visit the ED is if they have no other place to seek care when they have a medical problem. Access to healthcare is determined by many socioeconomic factors including having health insurance, income, occupation, hours or shifts worked, transportation, number and age of dependents and the patients’ social network. There is no standardized metric to measure access to healthcare. We attempted to control for healthcare access by adjusting for the patients’ age, insurance and estimated annual income in our models. The study identifies previous ED visits within the preceding year as a risk factor for ED visits within 30 days of OPAT initiation. Regardless of the underlying reasons driving ED visits, patients with a prior ED visit within the preceding year constitute an easily identifiable group that can be targeted for intervention designed to reduce ED visits during OPAT. Occlusion is the most common vascular access problem resulting in ED visits. The knowledge that ED visits for vascular access complications are much less likely to result in hospitalization than ED visits for worsening infection or non-OPAT-related reasons reinforces the strategy of managing occlusions at the site of outpatient care. In conclusion, about 5% of patients on OPAT have an ED visit within 30 days of initiation of OPAT. Compared with patients without, those with a prior ED visit within the preceding year have a 2.3-fold higher hazard of an ED visit. Only about half of the ED visits are for OPAT-related concerns, primarily vascular access problems. Patients who visit the ED for worsening infection are much more likely to be hospitalized than those who visit for vascular access complications. Acknowledgements We thank Robert Burton, Systems Analyst from the Cleveland Clinic ITD Analytics eResearch Department, for data retrieval from the EDV. Funding This study was conducted as part of our routine work. Transparency declarations None to declare. References 1 Tice AD , Rehm SJ , Dalovisio JR et al. Practice guidelines for outpatient parenteral antimicrobial therapy . Clin Infect Dis 2004 ; 38 : 1651 – 72 . Google Scholar CrossRef Search ADS PubMed 2 Lai A , Tran T , Nguyen HM et al. Outpatient parenteral antimicrobial therapy at large Veterans Administration medical center . Am J Manag Care 2013 ; 19 : e317 – 24 . Google Scholar PubMed 3 Paladino JA , Poretz D. Outpatient parenteral antimicrobial therapy today . Clin Infect Dis 2010 ; 51 Suppl 2: S198 – 208 . Google Scholar CrossRef Search ADS PubMed 4 Allison GM , Muldoon EG , Kent DM et al. Prediction model for 30-day hospital readmissions among patients discharged receiving outpatient parenteral antibiotic therapy . Clin Infect Dis 2014 ; 58 : 812 – 9 . Google Scholar CrossRef Search ADS PubMed 5 Means L , Bleasdale S , Sikka M et al. Predictors of hospital readmission in patients receiving outpatient parenteral antimicrobial therapy . Pharmacotherapy 2016 ; 36 : 934 – 9 . Google Scholar CrossRef Search ADS PubMed 6 Shrestha NK , Shrestha J , Everett A et al. Vascular access complications during outpatient parenteral antimicrobial therapy at home: a retrospective cohort study . J Antimicrob Chemother 2016 ; 71 : 506 – 12 . Google Scholar CrossRef Search ADS PubMed 7 Yan M , Elligsen M , Simor AE et al. Patient characteristics and outcomes of outpatient parenteral antimicrobial therapy: a retrospective study . Can J Infect Dis Med Microbiol 2016 ; 2016 : 8435257. Google Scholar PubMed 8 Gordon SM , Shrestha NK , Rehm SJ. Transitioning antimicrobial stewardship beyond the hospital: the Cleveland Clinic’s community-based parenteral anti-infective therapy (CoPAT) program . J Hosp Med 2011 ; 6 Suppl 1: S24 – 30 . Google Scholar CrossRef Search ADS PubMed 9 Internal Revenue Service . SOI Tax Stats—Individual Income Tax Statistics—2015 ZIP Code Data (SOI). 2017 . https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-statistics-2015-zip-code-data-soi. 10 Venables WN , Ripley BD. Modern Applied Statistics with S . New York, NY, USA : Springer , 2002 . Google Scholar CrossRef Search ADS 11 Fine JP , Gray RJ. A proportional hazards model for the subdistribution of a competing risk . J Am Stat Assoc 1999 ; 94 : 496 – 509 . Google Scholar CrossRef Search ADS 12 Gray B. cmprsk: Subdistribution Analysis of Competing Risks. R package. Vienna, Austria: R Foundation for Statistical Computing, 2015 . 13 Fox J , Monette G. Generalized collinearity diagnostics . J Am Stat Assoc 1992 ; 87 : 178 – 83 . Google Scholar CrossRef Search ADS 14 Fox J , Weisberg S. An R Companion to Applied Regression . Thousand Oaks, CA, USA : Sage , 2011 . 15 R Core Team . R: A Language and Environment for Statistical Computing . Vienna, Austria : R Foundation for Statistical Computing , 2017 . 16 Yoshida K , Bohn J. tableone: Create ‘Table 1’ to Describe Baseline Characteristics. R package. Vienna, Austria: R Foundation for Statistical Computing, 2017 . 17 Therneau TM. survival: A Package for Survival Analysis in S. R package. Vienna, Austria: R Foundation for Statistical Computing, 2015 . © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Antimicrobial Chemotherapy Oxford University Press

Emergency department visits during outpatient parenteral antimicrobial therapy: a retrospective cohort study

Loading next page...
 
/lp/ou_press/emergency-department-visits-during-outpatient-parenteral-antimicrobial-ltt0UnygDv
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please email: journals.permissions@oup.com.
ISSN
0305-7453
eISSN
1460-2091
D.O.I.
10.1093/jac/dky133
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives Patients receiving outpatient parenteral antimicrobial therapy (OPAT) may require emergency department (ED) visits to manage complications. This study’s purpose was to identify risk factors for ED visits during OPAT and risk factors for hospitalization among patients with ED visits. Methods All OPAT courses initiated between 1 January 2013 and 1 January 2017 at Cleveland Clinic were identified. The first OPAT course per patient was included. For these, ED visits within 30 days were identified. Reasons and risk factors for these visits were sought, as were risk factors for hospitalization among patients visiting the ED. Results Among 8263 patients on OPAT, 381 (4.6%) had at least one ED visit, an additional 1133 (14%) were hospitalized and an additional 50 (0.6%) died, within 30 days. One hundred and ninety-three ED visits (51%) were OPAT related. In a multivariable subdistribution proportional hazards competing risks regression model, prior ED visit (preceding year) was most strongly associated with ED visits (HR 2.29, 95% CI 1.76–2.98, P = 8.1 × 10−10). Eighty-five visits (22%) led to hospitalization. Compared with non-OPAT-related reasons, visits for vascular access complications were associated with lower odds of hospitalization (OR 0.36, 95% CI 0.14–0.83, P = 0.022) and visits for worsening infection with higher odds (OR 18.95, 95% CI 5.50–79.85, P = 1.2 × 10−7). Conclusions Compared with patients without, patients with prior ED visit have a 2.3-fold higher hazard of an ED visit within 30 days of OPAT initiation. Visits for worsening infection are much more likely to result in hospitalization than those for vascular access complications. Introduction Outpatient parenteral antimicrobial therapy (OPAT) has become a widely accepted form of therapy for patients who require intravenous antimicrobial therapy at home or other outpatient setting and it is estimated to be provided to 1 out of 1000 Americans every year.1 It has allowed for shorter hospitalizations, resulting in lower costs of healthcare.1,2 The initial model of OPAT was the visiting nurse model, where parenteral antimicrobials were infused in patients’ homes by visiting nurses. Over the years, many other models have been developed including the infusion centre model and the teach-and-train model.3 The visiting nurse model in the USA has evolved into primarily a self-administration model with periodic (typically once weekly) visiting nurse support. The goal of all these models is to allow safe administration of parenteral antimicrobials outside an inpatient hospital setting. Not every OPAT course concludes uneventfully. Unanticipated events that require attention include treatment failure, antimicrobial adverse events and vascular access complications. Some of these events lead to hospital admission. Previous hospitalization in the preceding 12 months, history of malignant lymphoma, longer planned OPAT duration, higher age, history of drug-resistant organisms, aminoglycoside use and lack of a primary care doctor have been identified as risk factors for hospital admission during OPAT.4,5 OPAT duration, female sex and injection drug use have been identified as risk factors for vascular access complications.6 Some patients on OPAT develop a need to visit an emergency department (ED). Such visits are unanticipated and unwelcome events and constitute utilization of an expensive component of the healthcare system. ED visits during OPAT have not received as much attention as hospitalizations. In one study that examined ED visits for patients on OPAT at a tertiary care centre in Canada, 43% of 104 patients discharged home on OPAT visited an ED within 60 days.7 An excess of ED visits stresses the healthcare system. Elucidation of the frequency of ED visits during OPAT will help quantify the magnitude of this stress. Identification of risk factors that result in ED visits may help identify opportunities for preventing them. No prior study has comprehensively examined demographic and clinical characteristics of OPAT and their association with ED visits. The purpose of this study was to describe ED visits during OPAT, to identify risk factors for ED visits and to identify risk factors for hospitalization among patients who visit the ED. Patients and methods Study design and setting The study was a retrospective cohort study conducted at Cleveland Clinic, a large multispecialty tertiary care referral centre in the USA. The Cleveland Clinic Health System (CCHS) includes the Cleveland Clinic Main Campus (a 1400 bed hospital), 10 regional hospitals in north-east Ohio and over 150 outpatient locations in northern Ohio, including 18 full-service family health centres and 21 express and urgent care sites. The entire health system uses a common electronic medical record (EMR), allowing for examination of patient care data throughout the system. Cleveland Clinic Main Campus has a large OPAT programme that has been in existence since 1979 and now encompasses ∼2500 OPAT courses a year. Initiation of OPAT at Cleveland Clinic requires evaluation by a Cleveland Clinic infectious disease attending physician.8 An electronic structured start-of-care form is completed by the infectious disease physician at the start of the OPAT course. The Cleveland Clinic OPAT registry identifies patients through the OPAT start-of-care forms. The common EMR allows for identification of ED visits not just to Cleveland Clinic Main Campus but to the entire CCHS. Screening, inclusion and exclusion criteria All OPAT courses for patients aged >18 years started between 1 January 2013 and 1 January 2017 at Cleveland Clinic Main Campus were identified from the Cleveland Clinic OPAT registry and screened for inclusion in this study. The first OPAT course per patient during the study period was included. OPAT courses that lasted <1 day were excluded. Data sources and data extraction Sources of data were the Cleveland Clinic OPAT registry and the Cleveland Clinic enterprise data vault (EDV). The Cleveland Clinic OPAT registry was used to identify OPAT episodes within the designated time frame, as well as the site of infection, antimicrobials, vascular access device, OPAT initiation date, the OPAT delivery site and planned duration of therapy. The comorbid conditions, residence ZIP code, county, health insurance and prior ED visits were obtained from the EDV. For the included patients, ED visits within the CCHS that occurred after the OPAT start date were searched for in the EDV and those that occurred within 30 days were identified. The reason for ED visit and patient disposition after the ED visit were ascertained by manual review of the EMR. For the sake of consistency, this was done by a single investigator. Study data were collected in a relational database (Microsoft® Access). Ethics The study was approved by the Cleveland Clinic institutional review board (IRB no. 16-1692). A waiver of the requirement for patient consent was granted. Definitions Age was defined as patient age on the OPAT start date. Comorbid conditions included diabetes mellitus, end-stage renal disease (on haemodialysis), COPD, liver cirrhosis and malignancy. Malignancy was defined as a history of haematological malignancy or solid tumour. The OPAT initiation site was ‘inpatient’ if the OPAT was begun during hospitalization and ‘outpatient’ if it was started in the outpatient setting. The OPAT delivery site could be home, residential facility or non-residential facility. Skilled nursing facilities and long-term acute care facilities were considered residential facilities. Infusion centres and dialysis centres were considered non-residential facilities. The county of residence was grouped into one of the following: Cuyahoga (county where Cleveland Clinic Main Campus is located), surrounding (county that has a border with Cuyahoga County), distant (a US county that does not border Cuyahoga County) and foreign (outside the USA). Medicare and Medicaid were considered government insurance. The mean annual household income for the year 2015 for the ZIP code in which the patient resided was taken to be the estimated patient income; it was calculated from publicly available data obtained from the Internal Revenue Service website.9 ED visits resulting from the infectious disease being treated, antimicrobial adverse events or vascular access problems were categorized as OPAT-related ED visits. Outcomes The primary outcome was days to ED visit. Death and hospital admission were recognized as competing outcomes that would preclude an ED visit. An ED visit that led to hospital admission was counted as an ED visit. Statistical analysis Baseline characteristics of OPAT courses with or without at least one ED visit were compared. Continuous variables were compared using Student’s t-test and categorical variables using the χ2 test. A frequency distribution of the reason for ED visit was tabulated. Risk factors for ED visits were identified in a multivariable Cox proportional hazards regression model by examining the association of relevant demographic and clinical factors, which were felt to possibly influence need for an ED visit, with the first ED visit per patient. The outcome examined was time to ED visit. The time-to-event variable was censored at 30 days. For patients who died or were subsequently hospitalized within 30 days, before they had a single ED visit, the variable was censored on the date of death or the date of hospitalization, respectively. Explanatory variables included in the initial model were patient age (decade of age), gender, calendar year, OPAT initiation site, OPAT delivery site, county of residence, insurance (government, private, international, uninsured), income quartile, diagnoses being treated with OPAT, parenteral antimicrobials administered, log anticipated OPAT duration and prior ED visit within the previous year. Variable selection was then done by backward stepwise elimination of the variable that caused the greatest reduction in the Akaike Information Criterion (AIC) at each step until the optimal model was obtained, using the stepAIC function in the R package MASS.10 Outliers were investigated for errors by examining case deletion residuals. The distribution of martingale residuals across values or levels of variables included in the final model was checked and the residuals were found to be evenly distributed. The proportional hazards assumption was checked using Schoenfeld residuals and the model was not in violation. To account for the competing risks of subsequent hospitalization or death, a subdistribution proportional hazards competing risks regression model was developed, with the variables in the final Cox proportional hazards model as the covariates, using the method of Fine and Gray.11,12 For the first ED visit per patient, for patients who did not have an intervening hospitalization, factors associated with hospital admission were examined in a multivariable logistic regression model including all the baseline variables as covariates. Variable selection was then done by backward stepwise selection to identify the optimal model based on identifying the model with the highest R2 for each given number of predictors and the model with the lowest AIC among these models.10 The final model was checked for multicollinearity using generalized variance inflation factors13,14 and there was no evidence of it. The final model was checked for interaction effects and none was found. Statistical analyses were done using R15 using the core functions in the base package and the packages tableone,16 survival,17 MASS,10 cmprsk12 and car.14 Results Between 1 January 2013 and 1 January 2017, 8263 patients were treated with a total of 11 391 OPAT courses. Of the 8263 OPAT courses included in the study (first OPAT course per patient), 416 (5%) had at least one ED visit within 30 days of initiation of OPAT, the first ED visit occurring a median of 10 days from the OPAT start date. Thirty-five had an intervening hospitalization between the OPAT start date and the first subsequent ED visit. Thus, 381 patients had at least one ED visit without an intervening hospitalization. Baseline characteristics of included patients The mean age at OPAT was 59 years (SD 16 years), 4737 (57%) were male and the median anticipated duration of OPAT was 20 days (IQR 11–34 days). The characteristics of the included patients are outlined in Table 1. Table 1. Baseline characteristics of included OPAT courses Characteristica ED visitb, N = 381 No ED visit, N = 7882 Pc Patient age (years), mean (SD) 57 (17) 59 (16) 0.004 Male, n (%) 189 (50) 4548 (58) 0.002 Year, n (%) <0.001  2013 215 (56) 2161 (27)  2014 76 (20) 1952 (25)  2015 45 (12) 1910 (24)  2016 45 (12) 1859 (24) Comorbid conditions, n (%)  diabetes mellitus 138 (36) 2817 (36) 0.891  end-stage renal disease 52 (14) 801 (10) 0.036  cirrhosis of liver 26 (7) 423 (5) 0.267  COPD 105 (28) 1615 (21) 0.001  malignancy 263 (69) 3964 (50) <0.001 Infection being treated, n (%)  abdominal 49 (13) 1014 (13) 1.000  cardiovascular 69 (18) 1544 (20) 0.519  CNS 23 (6) 538 (7) 0.622  genitourinary 34 (9) 491 (6) 0.046  head and neck 6 (2) 132 (2) 1.000  osteoarticular 104 (27) 2193 (28) 0.869  skin and skin structure 47 (12) 871 (11) 0.486  thoracic 23 (6) 625 (8) 0.213  vascular catheter infection/ unspecified bacteraemia 44 (12) 1014 (13) 0.501  other 6 (2) 89 (1) 0.582 Parenteral antimicrobial, n (%)  penicillins 127 (33) 2235 (28) 0.041  cephalosporins 66 (17) 1493 (19) 0.470  carbapenems 52 (14) 1232 (16) 0.332  glycopeptides 141 (37) 2822 (36) 0.671  lipopeptides 23 (6) 462 (6) 0.976  other antibacterials 23 (6) 395 (5) 0.440  antifungals 19 (5) 393 (5) 0.912  antivirals 7 (2) 201 (3) 0.484 Outpatient OPAT initiation, n (%) 18 (5) 419 (5) 0.699 Log anticipated OPAT duration (days), mean (SD) 2.79 (0.79) 2.88 (0.76) 0.024 OPAT delivery site, n (%) 0.152  home 220 (58) 4517 (57)  residential facility 142 (37) 3033 (39)  dialysis centre 19 (5) 274 (4)  infusion centre 0 (0) 58 (0.7) County of residence, n (%) <0.001  Cuyahoga 196 (51) 2407 (31)  surrounding 105 (28) 1517 (19)  distant 77 (20) 3890 (49)  foreign 3 (0.8) 68 (0.9) Insurance, n (%) <0.001  government 274 (72) 4919 (62)  private 79 (21) 2522 (32)  international 2 (0.5) 51 (0.6)  uninsured 26 (7) 390 (5) Income quartile (annual income), n (%) <0.001  USD 19 300–41 400 120 (32) 1838 (23)  USD 41 400–51 500 56 (15) 1919 (24)  USD 51 500–65 400 86 (23) 1889 (24)  USD 65 400–588 000 99 (26) 1915 (24)  unknown 20 (5) 321 (4) Vascular access device, n (%) 0.563  PICC 296 (78) 6025 (76)  Hohn 28 (7) 698 (9)  Hickman 16 (4) 336 (4)  indwelling port 11 (3) 296 (4)  midline 5 (1) 105 (1)  tunnelled dialysis catheter 21 (6) 303 (4)  other 4 (1) 119 (2) Infectious disease staff affiliationd, n (%) 0.435  bone and joint 111 (29) 2154 (27)  transplant 102 (27) 2106 (27)  neurological 59 (16) 1314 (17)  cardiac 47 (12) 1193 (15)  other 62 (16) 1115 (14) Prior ED visit within 1 year, n (%) 292 (77) 3640 (46) <0.001 Characteristica ED visitb, N = 381 No ED visit, N = 7882 Pc Patient age (years), mean (SD) 57 (17) 59 (16) 0.004 Male, n (%) 189 (50) 4548 (58) 0.002 Year, n (%) <0.001  2013 215 (56) 2161 (27)  2014 76 (20) 1952 (25)  2015 45 (12) 1910 (24)  2016 45 (12) 1859 (24) Comorbid conditions, n (%)  diabetes mellitus 138 (36) 2817 (36) 0.891  end-stage renal disease 52 (14) 801 (10) 0.036  cirrhosis of liver 26 (7) 423 (5) 0.267  COPD 105 (28) 1615 (21) 0.001  malignancy 263 (69) 3964 (50) <0.001 Infection being treated, n (%)  abdominal 49 (13) 1014 (13) 1.000  cardiovascular 69 (18) 1544 (20) 0.519  CNS 23 (6) 538 (7) 0.622  genitourinary 34 (9) 491 (6) 0.046  head and neck 6 (2) 132 (2) 1.000  osteoarticular 104 (27) 2193 (28) 0.869  skin and skin structure 47 (12) 871 (11) 0.486  thoracic 23 (6) 625 (8) 0.213  vascular catheter infection/ unspecified bacteraemia 44 (12) 1014 (13) 0.501  other 6 (2) 89 (1) 0.582 Parenteral antimicrobial, n (%)  penicillins 127 (33) 2235 (28) 0.041  cephalosporins 66 (17) 1493 (19) 0.470  carbapenems 52 (14) 1232 (16) 0.332  glycopeptides 141 (37) 2822 (36) 0.671  lipopeptides 23 (6) 462 (6) 0.976  other antibacterials 23 (6) 395 (5) 0.440  antifungals 19 (5) 393 (5) 0.912  antivirals 7 (2) 201 (3) 0.484 Outpatient OPAT initiation, n (%) 18 (5) 419 (5) 0.699 Log anticipated OPAT duration (days), mean (SD) 2.79 (0.79) 2.88 (0.76) 0.024 OPAT delivery site, n (%) 0.152  home 220 (58) 4517 (57)  residential facility 142 (37) 3033 (39)  dialysis centre 19 (5) 274 (4)  infusion centre 0 (0) 58 (0.7) County of residence, n (%) <0.001  Cuyahoga 196 (51) 2407 (31)  surrounding 105 (28) 1517 (19)  distant 77 (20) 3890 (49)  foreign 3 (0.8) 68 (0.9) Insurance, n (%) <0.001  government 274 (72) 4919 (62)  private 79 (21) 2522 (32)  international 2 (0.5) 51 (0.6)  uninsured 26 (7) 390 (5) Income quartile (annual income), n (%) <0.001  USD 19 300–41 400 120 (32) 1838 (23)  USD 41 400–51 500 56 (15) 1919 (24)  USD 51 500–65 400 86 (23) 1889 (24)  USD 65 400–588 000 99 (26) 1915 (24)  unknown 20 (5) 321 (4) Vascular access device, n (%) 0.563  PICC 296 (78) 6025 (76)  Hohn 28 (7) 698 (9)  Hickman 16 (4) 336 (4)  indwelling port 11 (3) 296 (4)  midline 5 (1) 105 (1)  tunnelled dialysis catheter 21 (6) 303 (4)  other 4 (1) 119 (2) Infectious disease staff affiliationd, n (%) 0.435  bone and joint 111 (29) 2154 (27)  transplant 102 (27) 2106 (27)  neurological 59 (16) 1314 (17)  cardiac 47 (12) 1193 (15)  other 62 (16) 1115 (14) Prior ED visit within 1 year, n (%) 292 (77) 3640 (46) <0.001 PICC, percutaneously inserted central catheter. a Expressed as mean (SD) for continuous variables and n (%) for categorical variables. b At least one ED visit within 30 days of initiation of OPAT. c Student’s t-test for continuous variables and the χ2 test for categorical variables. d Subspecialty affiliation of the infectious disease staff physician responsible for the OPAT course. Table 1. Baseline characteristics of included OPAT courses Characteristica ED visitb, N = 381 No ED visit, N = 7882 Pc Patient age (years), mean (SD) 57 (17) 59 (16) 0.004 Male, n (%) 189 (50) 4548 (58) 0.002 Year, n (%) <0.001  2013 215 (56) 2161 (27)  2014 76 (20) 1952 (25)  2015 45 (12) 1910 (24)  2016 45 (12) 1859 (24) Comorbid conditions, n (%)  diabetes mellitus 138 (36) 2817 (36) 0.891  end-stage renal disease 52 (14) 801 (10) 0.036  cirrhosis of liver 26 (7) 423 (5) 0.267  COPD 105 (28) 1615 (21) 0.001  malignancy 263 (69) 3964 (50) <0.001 Infection being treated, n (%)  abdominal 49 (13) 1014 (13) 1.000  cardiovascular 69 (18) 1544 (20) 0.519  CNS 23 (6) 538 (7) 0.622  genitourinary 34 (9) 491 (6) 0.046  head and neck 6 (2) 132 (2) 1.000  osteoarticular 104 (27) 2193 (28) 0.869  skin and skin structure 47 (12) 871 (11) 0.486  thoracic 23 (6) 625 (8) 0.213  vascular catheter infection/ unspecified bacteraemia 44 (12) 1014 (13) 0.501  other 6 (2) 89 (1) 0.582 Parenteral antimicrobial, n (%)  penicillins 127 (33) 2235 (28) 0.041  cephalosporins 66 (17) 1493 (19) 0.470  carbapenems 52 (14) 1232 (16) 0.332  glycopeptides 141 (37) 2822 (36) 0.671  lipopeptides 23 (6) 462 (6) 0.976  other antibacterials 23 (6) 395 (5) 0.440  antifungals 19 (5) 393 (5) 0.912  antivirals 7 (2) 201 (3) 0.484 Outpatient OPAT initiation, n (%) 18 (5) 419 (5) 0.699 Log anticipated OPAT duration (days), mean (SD) 2.79 (0.79) 2.88 (0.76) 0.024 OPAT delivery site, n (%) 0.152  home 220 (58) 4517 (57)  residential facility 142 (37) 3033 (39)  dialysis centre 19 (5) 274 (4)  infusion centre 0 (0) 58 (0.7) County of residence, n (%) <0.001  Cuyahoga 196 (51) 2407 (31)  surrounding 105 (28) 1517 (19)  distant 77 (20) 3890 (49)  foreign 3 (0.8) 68 (0.9) Insurance, n (%) <0.001  government 274 (72) 4919 (62)  private 79 (21) 2522 (32)  international 2 (0.5) 51 (0.6)  uninsured 26 (7) 390 (5) Income quartile (annual income), n (%) <0.001  USD 19 300–41 400 120 (32) 1838 (23)  USD 41 400–51 500 56 (15) 1919 (24)  USD 51 500–65 400 86 (23) 1889 (24)  USD 65 400–588 000 99 (26) 1915 (24)  unknown 20 (5) 321 (4) Vascular access device, n (%) 0.563  PICC 296 (78) 6025 (76)  Hohn 28 (7) 698 (9)  Hickman 16 (4) 336 (4)  indwelling port 11 (3) 296 (4)  midline 5 (1) 105 (1)  tunnelled dialysis catheter 21 (6) 303 (4)  other 4 (1) 119 (2) Infectious disease staff affiliationd, n (%) 0.435  bone and joint 111 (29) 2154 (27)  transplant 102 (27) 2106 (27)  neurological 59 (16) 1314 (17)  cardiac 47 (12) 1193 (15)  other 62 (16) 1115 (14) Prior ED visit within 1 year, n (%) 292 (77) 3640 (46) <0.001 Characteristica ED visitb, N = 381 No ED visit, N = 7882 Pc Patient age (years), mean (SD) 57 (17) 59 (16) 0.004 Male, n (%) 189 (50) 4548 (58) 0.002 Year, n (%) <0.001  2013 215 (56) 2161 (27)  2014 76 (20) 1952 (25)  2015 45 (12) 1910 (24)  2016 45 (12) 1859 (24) Comorbid conditions, n (%)  diabetes mellitus 138 (36) 2817 (36) 0.891  end-stage renal disease 52 (14) 801 (10) 0.036  cirrhosis of liver 26 (7) 423 (5) 0.267  COPD 105 (28) 1615 (21) 0.001  malignancy 263 (69) 3964 (50) <0.001 Infection being treated, n (%)  abdominal 49 (13) 1014 (13) 1.000  cardiovascular 69 (18) 1544 (20) 0.519  CNS 23 (6) 538 (7) 0.622  genitourinary 34 (9) 491 (6) 0.046  head and neck 6 (2) 132 (2) 1.000  osteoarticular 104 (27) 2193 (28) 0.869  skin and skin structure 47 (12) 871 (11) 0.486  thoracic 23 (6) 625 (8) 0.213  vascular catheter infection/ unspecified bacteraemia 44 (12) 1014 (13) 0.501  other 6 (2) 89 (1) 0.582 Parenteral antimicrobial, n (%)  penicillins 127 (33) 2235 (28) 0.041  cephalosporins 66 (17) 1493 (19) 0.470  carbapenems 52 (14) 1232 (16) 0.332  glycopeptides 141 (37) 2822 (36) 0.671  lipopeptides 23 (6) 462 (6) 0.976  other antibacterials 23 (6) 395 (5) 0.440  antifungals 19 (5) 393 (5) 0.912  antivirals 7 (2) 201 (3) 0.484 Outpatient OPAT initiation, n (%) 18 (5) 419 (5) 0.699 Log anticipated OPAT duration (days), mean (SD) 2.79 (0.79) 2.88 (0.76) 0.024 OPAT delivery site, n (%) 0.152  home 220 (58) 4517 (57)  residential facility 142 (37) 3033 (39)  dialysis centre 19 (5) 274 (4)  infusion centre 0 (0) 58 (0.7) County of residence, n (%) <0.001  Cuyahoga 196 (51) 2407 (31)  surrounding 105 (28) 1517 (19)  distant 77 (20) 3890 (49)  foreign 3 (0.8) 68 (0.9) Insurance, n (%) <0.001  government 274 (72) 4919 (62)  private 79 (21) 2522 (32)  international 2 (0.5) 51 (0.6)  uninsured 26 (7) 390 (5) Income quartile (annual income), n (%) <0.001  USD 19 300–41 400 120 (32) 1838 (23)  USD 41 400–51 500 56 (15) 1919 (24)  USD 51 500–65 400 86 (23) 1889 (24)  USD 65 400–588 000 99 (26) 1915 (24)  unknown 20 (5) 321 (4) Vascular access device, n (%) 0.563  PICC 296 (78) 6025 (76)  Hohn 28 (7) 698 (9)  Hickman 16 (4) 336 (4)  indwelling port 11 (3) 296 (4)  midline 5 (1) 105 (1)  tunnelled dialysis catheter 21 (6) 303 (4)  other 4 (1) 119 (2) Infectious disease staff affiliationd, n (%) 0.435  bone and joint 111 (29) 2154 (27)  transplant 102 (27) 2106 (27)  neurological 59 (16) 1314 (17)  cardiac 47 (12) 1193 (15)  other 62 (16) 1115 (14) Prior ED visit within 1 year, n (%) 292 (77) 3640 (46) <0.001 PICC, percutaneously inserted central catheter. a Expressed as mean (SD) for continuous variables and n (%) for categorical variables. b At least one ED visit within 30 days of initiation of OPAT. c Student’s t-test for continuous variables and the χ2 test for categorical variables. d Subspecialty affiliation of the infectious disease staff physician responsible for the OPAT course. Deaths, hospital admissions and ED visits Three hundred and eighty-one patients (4.6%) had at least one ED visit within 30 days of OPAT initiation in the absence of an intervening hospitalization. In addition, 50 patients (0.6%) died and 1133 (14%) were hospitalized within 30 days without an intervening ED visit. The latter were patients who were admitted to hospital directly, not through the ED. The cumulative probabilities of having these competing events (ED visit, hospital admission and death) over time are shown in Figure 1. Figure 1. View largeDownload slide Cumulative occurrence of competing events (ED visit, hospital admission and death) in the 30 days following initiation of OPAT. Figure 1. View largeDownload slide Cumulative occurrence of competing events (ED visit, hospital admission and death) in the 30 days following initiation of OPAT. Description of ED visits Of the 381 ED visits (first ED visit for patients who had at least one ED visit without an intervening hospitalization), 193 (51%) were OPAT related. The most common visit reason was vascular access complication, which accounted for 104 ED visits (54% of OPAT-related ED visits, 27% of all ED visits). The most common vascular access complications were occlusion and dislodgement, which accounted for 53 and 19 ED visits, respectively. The proportion of OPAT courses with at least one ED visit were 9.1%, 3.8%, 2.3% and 2.4%, respectively, for the years 2013, 2014, 2015 and 2016. Reasons for ED visit are outlined in Table 2. Table 2. Frequency distribution of reason for ED visit ED visit reason n Percentage of ED visits OPAT related  worsening infectiona 18 5  antimicrobial adverse event 14 4  vascular access complication 104 27  symptom associated with OPAT diagnosisb 57 15 Non-OPAT related  exacerbation of a chronic condition 43 11  complication of a chronic condition 24 6  new non-OPAT problem 121 32 ED visit reason n Percentage of ED visits OPAT related  worsening infectiona 18 5  antimicrobial adverse event 14 4  vascular access complication 104 27  symptom associated with OPAT diagnosisb 57 15 Non-OPAT related  exacerbation of a chronic condition 43 11  complication of a chronic condition 24 6  new non-OPAT problem 121 32 a Symptom(s) deemed to be from worsening infection after clinical evaluation. b Symptoms related to the infection being treated but deemed not to be worsening infection after clinical evaluation. Table 2. Frequency distribution of reason for ED visit ED visit reason n Percentage of ED visits OPAT related  worsening infectiona 18 5  antimicrobial adverse event 14 4  vascular access complication 104 27  symptom associated with OPAT diagnosisb 57 15 Non-OPAT related  exacerbation of a chronic condition 43 11  complication of a chronic condition 24 6  new non-OPAT problem 121 32 ED visit reason n Percentage of ED visits OPAT related  worsening infectiona 18 5  antimicrobial adverse event 14 4  vascular access complication 104 27  symptom associated with OPAT diagnosisb 57 15 Non-OPAT related  exacerbation of a chronic condition 43 11  complication of a chronic condition 24 6  new non-OPAT problem 121 32 a Symptom(s) deemed to be from worsening infection after clinical evaluation. b Symptoms related to the infection being treated but deemed not to be worsening infection after clinical evaluation. Risk factors for ED visits Death and hospital admission were recognized as competing outcomes for ED visit and this was accounted for in the analysis. In a multivariable subdistribution proportional hazards competing risks regression model, the variables that were significantly associated with having an ED visit are shown in Table 3. Notably, patients with a prior ED visit within the preceding year had a 2.3-fold higher hazard of having an ED visit within 30 days of starting OPAT than patients who had not been to an ED in the preceding year (HR 2.29, 95% CI 1.76–2.98, P = 8.1 × 10−10). Other factors associated with a significantly higher hazard of ED visits were female sex (HR 1.25, 95% CI 1.02–1.53, P = 0.032), malignancy (HR 1.49, 95% CI 1.19–1.87, P = 6.1 × 10−4) and treatment with a parenteral penicillin (HR 1.31, 95% CI 1.06–1.61, P = 0.014). Factors associated with a significantly lower hazard of ED visits were higher age (HR 0.89 per decade of age, 95% CI 0.83–0.95, P = 4.0 × 10−4), distant county as opposed to Cuyahoga County (HR 0.42, 95% CI 0.31–0.56, P = 8.5 × 10−9) and private insurance as opposed to government insurance (HR 0.59, 95% CI 0.46–0.77, P = 6.3 × 10−5). The hazard for ED visit was lower for each subsequent year compared with the year 2013. Table 3. Factors significantly associated with ED visits, for patients on OPAT, in a multivariable subdistribution proportional hazards competing risks regression model Factora HR 95% CI P Age (decade) 0.89 0.83–0.95 4.0 × 10−4 Female 1.25 1.02–1.53 0.032 Malignancy 1.49 1.19–1.87 6.1 × 10−4 Treatment with a parenteral penicillin 1.31 1.06–1.61 0.014 County (reference: Cuyahoga)  surrounding 1.04 0.81–1.32 0.78  distant 0.42 0.31–0.56 8.5 × 10−9  foreign 0.87 0.22–3.39 0.84 Insurance (reference: government)  private 0.59 0.46–0.77 6.3 × 10−5  international 0.79 0.16–3.92 0.77  uninsured 1.16 0.75–1.80 0.50 Year (reference: 2013)  2014 0.42 0.32–0.54 6.8 × 10−11  2015 0.25 0.18–0.35 <2.2 × 10−16  2016 0.27 0.20–0.37 1.6 × 10−15 Prior ED visitb 2.29 1.76–2.98 8.1 × 10−10 Factora HR 95% CI P Age (decade) 0.89 0.83–0.95 4.0 × 10−4 Female 1.25 1.02–1.53 0.032 Malignancy 1.49 1.19–1.87 6.1 × 10−4 Treatment with a parenteral penicillin 1.31 1.06–1.61 0.014 County (reference: Cuyahoga)  surrounding 1.04 0.81–1.32 0.78  distant 0.42 0.31–0.56 8.5 × 10−9  foreign 0.87 0.22–3.39 0.84 Insurance (reference: government)  private 0.59 0.46–0.77 6.3 × 10−5  international 0.79 0.16–3.92 0.77  uninsured 1.16 0.75–1.80 0.50 Year (reference: 2013)  2014 0.42 0.32–0.54 6.8 × 10−11  2015 0.25 0.18–0.35 <2.2 × 10−16  2016 0.27 0.20–0.37 1.6 × 10−15 Prior ED visitb 2.29 1.76–2.98 8.1 × 10−10 a COPD (P = 0.27), outpatient initiation (P = 0.27) and log anticipated OPAT duration (P = 0.45) were also included as variables in the final model but were not statistically significant. b Prior ED visit within the preceding year. Table 3. Factors significantly associated with ED visits, for patients on OPAT, in a multivariable subdistribution proportional hazards competing risks regression model Factora HR 95% CI P Age (decade) 0.89 0.83–0.95 4.0 × 10−4 Female 1.25 1.02–1.53 0.032 Malignancy 1.49 1.19–1.87 6.1 × 10−4 Treatment with a parenteral penicillin 1.31 1.06–1.61 0.014 County (reference: Cuyahoga)  surrounding 1.04 0.81–1.32 0.78  distant 0.42 0.31–0.56 8.5 × 10−9  foreign 0.87 0.22–3.39 0.84 Insurance (reference: government)  private 0.59 0.46–0.77 6.3 × 10−5  international 0.79 0.16–3.92 0.77  uninsured 1.16 0.75–1.80 0.50 Year (reference: 2013)  2014 0.42 0.32–0.54 6.8 × 10−11  2015 0.25 0.18–0.35 <2.2 × 10−16  2016 0.27 0.20–0.37 1.6 × 10−15 Prior ED visitb 2.29 1.76–2.98 8.1 × 10−10 Factora HR 95% CI P Age (decade) 0.89 0.83–0.95 4.0 × 10−4 Female 1.25 1.02–1.53 0.032 Malignancy 1.49 1.19–1.87 6.1 × 10−4 Treatment with a parenteral penicillin 1.31 1.06–1.61 0.014 County (reference: Cuyahoga)  surrounding 1.04 0.81–1.32 0.78  distant 0.42 0.31–0.56 8.5 × 10−9  foreign 0.87 0.22–3.39 0.84 Insurance (reference: government)  private 0.59 0.46–0.77 6.3 × 10−5  international 0.79 0.16–3.92 0.77  uninsured 1.16 0.75–1.80 0.50 Year (reference: 2013)  2014 0.42 0.32–0.54 6.8 × 10−11  2015 0.25 0.18–0.35 <2.2 × 10−16  2016 0.27 0.20–0.37 1.6 × 10−15 Prior ED visitb 2.29 1.76–2.98 8.1 × 10−10 a COPD (P = 0.27), outpatient initiation (P = 0.27) and log anticipated OPAT duration (P = 0.45) were also included as variables in the final model but were not statistically significant. b Prior ED visit within the preceding year. Factors associated with hospital admission for patients who had ED visits Among the 381 ED visits, 85 (22%) resulted in hospital admission. The factors that were independently associated with hospital admission are outlined in Table 4. Among patients who visited the ED within 30 days of starting OPAT, patients on treatment with a parenteral antiviral agent had 33-fold higher odds of being admitted than patients not on parenteral antiviral agents (OR 33.37, 95% CI 3.98–384.71, P = 0.002). Compared with patients who visited the ED for non-OPAT-related reasons, those presenting with worsening infection had 19-fold higher odds (OR 18.95, 95% CI 5.50–79.85, P = 1.2 × 10−7) and those presenting with vascular access problems had 3-fold lower odds (OR 0.36, 95% CI 0.14–0.83, P = 0.022) of being admitted to hospital. The decision to admit to hospital from the ED was also influenced significantly by which ED the patient presented to. Table 4. Significant associations with hospital admission, for ED visits, in a multivariable logistic regression model Factora OR 95% CI P Log anticipated OPAT duration 0.66 0.45–0.96 0.031 Parenteral antiviral treatment 33.37 3.98–384.71 0.002 ED (reference: Cleveland Clinic Main Campus)b  Ashtabula (NT and 80 kmc) 3.64 1.06–11.64 0.033  Avon (NT and 30 kmc) 9.03 2.96–28.02 1.1 × 10−4  Fairview (T and 18 kmc) 3.35 1.12–9.44 0.024  Marymount (NT and 9 kmc) 3.82 1.01–12.96 0.037  Medina (NT and 44 kmc) 9.63 3.96–24.29 8.6 × 10−7  Twinsburg (NT and 27 kmc) 9.32 2.79–31.27 2.6 × 10−4 ED visit reason (reference: non-OPAT reason)d  vascular access problem 0.36 0.14–0.83 0.022  worsening infection 18.95 5.50–79.85 1.2 × 10−7 Factora OR 95% CI P Log anticipated OPAT duration 0.66 0.45–0.96 0.031 Parenteral antiviral treatment 33.37 3.98–384.71 0.002 ED (reference: Cleveland Clinic Main Campus)b  Ashtabula (NT and 80 kmc) 3.64 1.06–11.64 0.033  Avon (NT and 30 kmc) 9.03 2.96–28.02 1.1 × 10−4  Fairview (T and 18 kmc) 3.35 1.12–9.44 0.024  Marymount (NT and 9 kmc) 3.82 1.01–12.96 0.037  Medina (NT and 44 kmc) 9.63 3.96–24.29 8.6 × 10−7  Twinsburg (NT and 27 kmc) 9.32 2.79–31.27 2.6 × 10−4 ED visit reason (reference: non-OPAT reason)d  vascular access problem 0.36 0.14–0.83 0.022  worsening infection 18.95 5.50–79.85 1.2 × 10−7 a Age (P = 0.06), diabetes mellitus (P = 0.10) and OPAT site (P = 0.06) were also included as variables in the final model but were not statistically significant. b Lakewood (P = 0.38), Lutheran (P = 0.29) and other EDs (P = 0.29) not significant. c Institution teaching status (T = teaching and NT = non-teaching) and distance from Cleveland Clinic Main Campus. d Antimicrobial adverse event (P = 0.49) and symptoms associated with OPAT diagnosis (P = 0.16) not significant. Table 4. Significant associations with hospital admission, for ED visits, in a multivariable logistic regression model Factora OR 95% CI P Log anticipated OPAT duration 0.66 0.45–0.96 0.031 Parenteral antiviral treatment 33.37 3.98–384.71 0.002 ED (reference: Cleveland Clinic Main Campus)b  Ashtabula (NT and 80 kmc) 3.64 1.06–11.64 0.033  Avon (NT and 30 kmc) 9.03 2.96–28.02 1.1 × 10−4  Fairview (T and 18 kmc) 3.35 1.12–9.44 0.024  Marymount (NT and 9 kmc) 3.82 1.01–12.96 0.037  Medina (NT and 44 kmc) 9.63 3.96–24.29 8.6 × 10−7  Twinsburg (NT and 27 kmc) 9.32 2.79–31.27 2.6 × 10−4 ED visit reason (reference: non-OPAT reason)d  vascular access problem 0.36 0.14–0.83 0.022  worsening infection 18.95 5.50–79.85 1.2 × 10−7 Factora OR 95% CI P Log anticipated OPAT duration 0.66 0.45–0.96 0.031 Parenteral antiviral treatment 33.37 3.98–384.71 0.002 ED (reference: Cleveland Clinic Main Campus)b  Ashtabula (NT and 80 kmc) 3.64 1.06–11.64 0.033  Avon (NT and 30 kmc) 9.03 2.96–28.02 1.1 × 10−4  Fairview (T and 18 kmc) 3.35 1.12–9.44 0.024  Marymount (NT and 9 kmc) 3.82 1.01–12.96 0.037  Medina (NT and 44 kmc) 9.63 3.96–24.29 8.6 × 10−7  Twinsburg (NT and 27 kmc) 9.32 2.79–31.27 2.6 × 10−4 ED visit reason (reference: non-OPAT reason)d  vascular access problem 0.36 0.14–0.83 0.022  worsening infection 18.95 5.50–79.85 1.2 × 10−7 a Age (P = 0.06), diabetes mellitus (P = 0.10) and OPAT site (P = 0.06) were also included as variables in the final model but were not statistically significant. b Lakewood (P = 0.38), Lutheran (P = 0.29) and other EDs (P = 0.29) not significant. c Institution teaching status (T = teaching and NT = non-teaching) and distance from Cleveland Clinic Main Campus. d Antimicrobial adverse event (P = 0.49) and symptoms associated with OPAT diagnosis (P = 0.16) not significant. Discussion This study finds that, excluding patients who were admitted directly to hospital without passing through the ED, about 5% of patients on OPAT have at least one ED visit within the 30 days following initiation of OPAT. Patients with an ED visit within the previous year have a 2.3-fold higher hazard of having an ED visit than patients without an ED visit in the preceding year. This is probably partly a marker of access to care. Some patients have no other recourse than to visit an ED when faced with a medical problem that requires attention. It is also possible that some patients with repeated ED visits are those who have medical conditions that lead to more medical emergencies. About half of the ED visits are for non-OPAT-related reasons. This is a reflection of the underlying comorbidity burden in many patients who develop a condition that requires parenteral antimicrobial therapy. The study also shows that the decision to admit to hospital from the ED varies significantly by the reason for the ED visit. Patients visiting the ED for vascular access problems are much less likely to be admitted to hospital than patients visiting for worsening infection or non-OPAT-related reasons. This is probably because vascular access problems that bring patients to the ED can usually be addressed without having to admit the patient to the hospital. The proportion of patients with an ED visit in our study (5% within 30 days) is much lower than was noted in a former study in Canada (43% within 60 days).7 This difference can be explained by the difference in the study setting and the fact that duration of follow-up in the Canadian study was twice as long. The Canadian study was done in a centre without a formal OPAT programme and only half of the patients had an infectious disease referral. Our study was done in an institution with a well-organized OPAT programme where every patient is seen by an infectious disease physician. A notable finding in our study is a drop in the proportion of patients with ED visits from 9% in 2013 to ∼2.4% in 2015 and 2016. One of the factors that might have contributed to this decline was an initiative taken by the most frequently used home healthcare provider to manage catheter occlusions by administering recombinant tissue plasminogen activator in patients’ homes instead of sending them to the ED, a practice that was also soon adopted by other agencies. But this alone would not be an adequate explanation for this reduction. It is very likely that concomitant efforts by the institution to reduce hospital readmissions (a response to the introduction of financial penalties for 30 day readmissions by government payors) also inadvertently resulted in reduction of ED visits. The biggest strength of this study is the large sample size. It allowed for use of more comprehensive models including many relevant variables that would not have been possible with a smaller study sample. Our OPAT programme and OPAT registry allowed for an accurate identification of all OPAT courses within our hospital.8 Including all ED visits across the health system allowed for a more comprehensive picture of the ED visit burden than would have been possible if we had only examined ED visits in our own hospital. In a rapidly changing healthcare environment, incentives to visit or avoid the ED change with time and could influence the results of any study examining ED visits. To control for this factor we included calendar year as a covariate in our models, thereby adjusting for policy or process factors that might have changed over time. The study has its limitations. We were only able to identify ED visits within the CCHS. There were certainly ED visits that occurred outside the CCHS, particularly for patients who lived at great distances from Cleveland Clinic. Cleveland Clinic Hospital is located in Cuyahoga County and all the EDs of the CCHS are located in Cuyahoga County or the surrounding counties. To control for the likelihood of visiting one of the EDs within the CCHS, we adjusted for the patients’ county of residence in our models to account for the fact that patients who lived closer to the hospitals of the CCHS were more likely to visit one of the EDs associated with the CCHS. We also recognized that an important reason for patients to visit the ED is if they have no other place to seek care when they have a medical problem. Access to healthcare is determined by many socioeconomic factors including having health insurance, income, occupation, hours or shifts worked, transportation, number and age of dependents and the patients’ social network. There is no standardized metric to measure access to healthcare. We attempted to control for healthcare access by adjusting for the patients’ age, insurance and estimated annual income in our models. The study identifies previous ED visits within the preceding year as a risk factor for ED visits within 30 days of OPAT initiation. Regardless of the underlying reasons driving ED visits, patients with a prior ED visit within the preceding year constitute an easily identifiable group that can be targeted for intervention designed to reduce ED visits during OPAT. Occlusion is the most common vascular access problem resulting in ED visits. The knowledge that ED visits for vascular access complications are much less likely to result in hospitalization than ED visits for worsening infection or non-OPAT-related reasons reinforces the strategy of managing occlusions at the site of outpatient care. In conclusion, about 5% of patients on OPAT have an ED visit within 30 days of initiation of OPAT. Compared with patients without, those with a prior ED visit within the preceding year have a 2.3-fold higher hazard of an ED visit. Only about half of the ED visits are for OPAT-related concerns, primarily vascular access problems. Patients who visit the ED for worsening infection are much more likely to be hospitalized than those who visit for vascular access complications. Acknowledgements We thank Robert Burton, Systems Analyst from the Cleveland Clinic ITD Analytics eResearch Department, for data retrieval from the EDV. Funding This study was conducted as part of our routine work. Transparency declarations None to declare. References 1 Tice AD , Rehm SJ , Dalovisio JR et al. Practice guidelines for outpatient parenteral antimicrobial therapy . Clin Infect Dis 2004 ; 38 : 1651 – 72 . Google Scholar CrossRef Search ADS PubMed 2 Lai A , Tran T , Nguyen HM et al. Outpatient parenteral antimicrobial therapy at large Veterans Administration medical center . Am J Manag Care 2013 ; 19 : e317 – 24 . Google Scholar PubMed 3 Paladino JA , Poretz D. Outpatient parenteral antimicrobial therapy today . Clin Infect Dis 2010 ; 51 Suppl 2: S198 – 208 . Google Scholar CrossRef Search ADS PubMed 4 Allison GM , Muldoon EG , Kent DM et al. Prediction model for 30-day hospital readmissions among patients discharged receiving outpatient parenteral antibiotic therapy . Clin Infect Dis 2014 ; 58 : 812 – 9 . Google Scholar CrossRef Search ADS PubMed 5 Means L , Bleasdale S , Sikka M et al. Predictors of hospital readmission in patients receiving outpatient parenteral antimicrobial therapy . Pharmacotherapy 2016 ; 36 : 934 – 9 . Google Scholar CrossRef Search ADS PubMed 6 Shrestha NK , Shrestha J , Everett A et al. Vascular access complications during outpatient parenteral antimicrobial therapy at home: a retrospective cohort study . J Antimicrob Chemother 2016 ; 71 : 506 – 12 . Google Scholar CrossRef Search ADS PubMed 7 Yan M , Elligsen M , Simor AE et al. Patient characteristics and outcomes of outpatient parenteral antimicrobial therapy: a retrospective study . Can J Infect Dis Med Microbiol 2016 ; 2016 : 8435257. Google Scholar PubMed 8 Gordon SM , Shrestha NK , Rehm SJ. Transitioning antimicrobial stewardship beyond the hospital: the Cleveland Clinic’s community-based parenteral anti-infective therapy (CoPAT) program . J Hosp Med 2011 ; 6 Suppl 1: S24 – 30 . Google Scholar CrossRef Search ADS PubMed 9 Internal Revenue Service . SOI Tax Stats—Individual Income Tax Statistics—2015 ZIP Code Data (SOI). 2017 . https://www.irs.gov/statistics/soi-tax-stats-individual-income-tax-statistics-2015-zip-code-data-soi. 10 Venables WN , Ripley BD. Modern Applied Statistics with S . New York, NY, USA : Springer , 2002 . Google Scholar CrossRef Search ADS 11 Fine JP , Gray RJ. A proportional hazards model for the subdistribution of a competing risk . J Am Stat Assoc 1999 ; 94 : 496 – 509 . Google Scholar CrossRef Search ADS 12 Gray B. cmprsk: Subdistribution Analysis of Competing Risks. R package. Vienna, Austria: R Foundation for Statistical Computing, 2015 . 13 Fox J , Monette G. Generalized collinearity diagnostics . J Am Stat Assoc 1992 ; 87 : 178 – 83 . Google Scholar CrossRef Search ADS 14 Fox J , Weisberg S. An R Companion to Applied Regression . Thousand Oaks, CA, USA : Sage , 2011 . 15 R Core Team . R: A Language and Environment for Statistical Computing . Vienna, Austria : R Foundation for Statistical Computing , 2017 . 16 Yoshida K , Bohn J. tableone: Create ‘Table 1’ to Describe Baseline Characteristics. R package. Vienna, Austria: R Foundation for Statistical Computing, 2017 . 17 Therneau TM. survival: A Package for Survival Analysis in S. R package. Vienna, Austria: R Foundation for Statistical Computing, 2015 . © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For permissions, please email: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Journal of Antimicrobial ChemotherapyOxford University Press

Published: Apr 18, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off