Advertisement

Evaluating the post-discharge cost of healthcare-associated infection in NHS Scotland

      Summary

      Background

      Whereas the cost burden of healthcare-associated infection (HAI) extends beyond the inpatient stay into the post-discharge period, few studies have focused on post-discharge costs.

      Aim

      To investigate the impact of all types of HAI on the magnitude and distribution of post-discharge costs observed in acute and community services for patients who developed HAI during their inpatient stay.

      Methods

      Using data from the Evaluation of Cost of Nosocomial Infection (ECONI) study and regression methods, this study identifies the marginal effect of HAI on the 90-daypost-discharge resource use and costs. To calculate monetary values, unit costs were applied to estimates of excess resource use per case of HAI.

      Findings

      Post-discharge costs increase inpatient HAI costs by 36%, with an annual national cost of £10,832,437. The total extra cost per patient with HAI was £1,457 (95% confidence interval: 1,004–4,244) in the 90 days post discharge. Patients with HAI had longer LOS if they were readmitted and were prescribed more antibiotics in the community. The results suggest that HAI did not have an impact on the number of readmissions or repeat surgeries within 90 days of discharge. The majority (95%) of the excess costs was on acute care services after readmission. Bloodstream infection, gastrointestinal infection, and pneumonia had the biggest impact on post-discharge cost.

      Conclusion

      HAI increases costs and antibiotic consumption in the post-discharge period. Economic evaluations of IPC studies should incorporate post-discharge costs. These findings can be used nationally and internationally to support decision-making on the impact of IPC interventions.

      Keywords

      Introduction

      Healthcare-associated infections (HAIs) increase the cost of healthcare and impose additional morbidity and risk of mortality [
      World Health Organization
      Report on the burden of endemic health care-associated infection worldwide.
      ]. The cost burden extends beyond the inpatient stay into the post-discharge period, yet a 2005 review showed that most published studies focus on inpatient costs only, particularly additional length of stay (LOS), and ignore the post-discharge period [
      • Stone P.W.
      • Braccia D.
      • Larson E.
      Systematic review of economic analyses of health care-associated infections.
      ]. Previous research has mostly focused on the post-discharge incidence of surgical site infection (SSI) and associated costs [
      • Graves N.
      • Halton K.
      • Curtis M.
      • Doidge S.
      • Lairson D.
      • McLaws M.
      • et al.
      Costs of surgical site infections that appear after hospital discharge.
      ,
      • Perencevich E.N.
      • Sands K.E.
      • Cosgrove S.E.
      • Guadagnoli E.
      • Meara E.
      • Platt R.
      Health and economic impact of surgical site infections diagnosed after hospital discharge.
      ,
      • Limon E.
      • Shaw E.
      • Badia J.M.
      • Piriz M.
      • Escofet R.
      • Gudiol F.
      • et al.
      Post-discharge surgical site infections after uncomplicated elective colorectal surgery: impact and risk factors. The experience of the VINCat Program.
      ]. Reporting post-discharge impact of HAI requires long-term follow-up of patients with, and without, HAI once they have left hospital. Recording healthcare utilization by these patients, which can be resource intensive to collate, is further complicated by the requirement to adjust for comorbidities seen within this patient population. The challenges and variability of estimates have been well described [
      • Roberts R.R.
      • Scott 2nd, R.D.
      • Hota B.
      • Kampe L.M.
      • Abbasi F.
      • Schabowski S.
      • et al.
      Costs attributable to healthcare-acquired infection in hospitalized adults and a comparison of economic methods.
      ]. Plowman et al. reported similar post-discharge resource use between HAI and non-HAI groups [
      • Plowman R.
      • Graves N.
      • Griffin M.A.
      • Roberts J.A.
      • Swan A.V.
      • Cookson B.
      • et al.
      The rate and cost of hospital-acquired infections occurring in patients admitted to selected specialties of a district general hospital in England and the national burden imposed.
      ,
      • Plowman R.
      • Graves N.
      • Griffin M.A.S.
      • Roberts J.
      • Swan A.V.
      • Cookson B.
      • et al.
      ]. Other studies have shown an increase in hospital readmissions in HAI patients when compared to patients without HAI [
      • Chopra T.
      • Neelakanta A.
      • Dombecki C.
      • Awali R.A.
      • Sharma S.
      • Kaye K.S.
      • et al.
      Burden of Clostridium difficile infection on hospital readmissions and its potential impact under the Hospital Readmission Reduction Program.
      ,
      • de Souza E.C.O.
      • Ferreira Filho S.P.
      • Gervatauskas K.M.
      • Fortaleza C.M.C.B.
      Post-discharge impact of healthcare-associated infections in a developing country: a cohort study.
      ,
      • Khong C.J.
      • Baggs J.
      • Kleinbaum D.
      • Cochran R.
      • Jernigan J.A.
      The likelihood of hospital readmission among patients with hospital-onset central line-associated bloodstream infections.
      ,
      • Stevens V.
      • Geiger K.
      • Concannon C.
      • Nelson R.E.
      • Brown J.
      • Dumyati G.
      Inpatient costs, mortality and 30-dayre-admission in patients with central-line-associated bloodstream infections.
      ,
      • Nelson R.E.
      • Jones M.
      • Liu C.F.
      • Samore M.H.
      • Evans M.E.
      • Graves N.
      • et al.
      The impact of healthcare-associated methicillin-resistant Staphylococcus aureus infections on post-discharge healthcare costs and utilization.
      ]. One review in the USA estimated that between 9% and 13% of the total health cost of HAI occurs post discharge and highlighted the wide range of methodologies, different HAI types, settings and differences in reporting [
      • Marchetti A.
      • Rossiter R.
      Economic burden of healthcare-associated infection in US acute care hospitals: societal perspective.
      ]. These create challenges when interpreting the findings of the few studies that include a perspective broader than acute hospitals.
      Reliable estimates on the total burden of HAI are critical for assessing the cost-effectiveness of infection prevention and control (IPC) programmes [
      • Graves N.
      • Harbarth S.
      • Beyersmann J.
      • Barnett A.
      • Halton K.
      • Cooper B.
      Estimating the cost of health care-associated infections: mind your p’s and q’s.
      ]. This requires estimates of resource use incurred during both the inpatient and post-discharge phases. With a growing focus on minimizing the duration of hospital stays there is an increasing requirement for acute and community services to be involved in patient care after a patient has been discharged from hospital. However, little is known about the impact of all types of HAI on the level and distribution of post-discharge resource use among acute and community services. The aim of this study was to investigate the impact of all types of HAI on the magnitude and distribution of post-discharge costs observed in acute and community services for patients who developed HAI during their inpatient stay. As such, the study provides information that can be used in the economic evaluation of IPC measures, as the financial benefits of preventing HAI in hospital may extend after the inpatient stay.

      Methods

      The analysis focuses on resource use and resulting costs that fall directly on the health service in the 90 days after patients are discharged from hospital. A 90-daypost-discharge period was chosen to fully capture the impact on resource use that can be attributed to the preceding hospital stay.

      Incidence of HAI

      The Evaluation of Cost of Nosocomial Infection (ECONI) study was a two-centre, prospective observational HAI incidence study with record linkage. The study was undertaken in one Scottish NHS teaching hospital and one general hospital, which were selected as being broadly representative of other acute hospitals of their type in Scotland in terms of patient specialties, HAI prevalence, and patient mix [
      Health Protection Scotland
      Scottish national point prevalence survey of healthcare associated infection and antimicrobial prescribing 2016.
      ]. Data collection began in April 2018 and continued for one calendar year. All adult inpatients admitted overnight were included within an incidence cohort. HAI cases were defined using the ECDC epidemiological case definitions [
      • Stewart S.
      • Robertson C.
      • Manoukian S.
      • Haahr L.
      • Mason H.
      • McFarland A.
      • et al.
      How do we evaluate the cost of nosocomial infection? The ECONI protocol: an incidence study with nested case–control evaluating cost and quality of life.
      ,
      European Centre for Disease Prevention and Control
      Point prevalence survey of healthcare-associated infections and antimicrobial use in European acute care hospitals – protocol version 5.3.
      ]. National estimates of incidence are reported elsewhere within each specialty group and hospital type in 2018/19 [
      • Stewart S.
      • Robertson C.
      • Pan J.
      • Kennedy S.
      • Dancer S.
      • Haahr L.
      • et al.
      Epidemiology of healthcare-associated infection reported from a hospital-wide incidence study: considerations for infection prevention and control planning.
      ].

      Identification of excess post-discharge resource use

      All adult overnight admissions to the study hospitals, including HAI cases and non-cases, were linked to NHS Scotland administrative electronic databases that contain information on discharge specialty, reason for admission, post-discharge episodes of care, post-discharge healthcare contacts, surgical interventions, and post-admission prescribing [
      Health and Social Care Information Centre
      OPCS classification of interventions and procedures version 4.7 volume I – tabular list.
      ,
      Information Services Division (ISD) Scotland
      SMR01 – general/acute inpatient and day case.
      ,
      Information Services Division (ISD) Scotland
      SMR00 – outpatient attendance.
      ,
      Information Services Division (ISD) Scotland
      Prescribing information system for Scotland (PRISMS).
      ].
      During the study there were 99,018 admissions to the hospitals. This analysis includes 664 patients who developed HAI and 43,841 patients with no HAI (see Appendix Figure A1). Patients who did not have a complete 90 days follow-up were excluded from the analysis. Patients who died within 90 days post discharge were excluded (2604 people died in hospital including 158 with HAI and 2374 people died post discharge including 57 with HAI). Patients were also excluded if they did not have any post-discharge follow-up information available (in total 2631 people 161 of which had HAI), plus a small number of patients who had been discharged from hospital for less than 90 days (31 people all non-HAI).
      Regression analysis at the patient level was employed to identify the marginal effect of HAI on post-discharge resource use also referred to as patient outcomes. Patient outcomes in this study included readmissions to acute care hospitals, outpatient visits, community prescribing and general practitioner (GP) prescribing costs. In the regression analysis, one admission or index inpatient stay per patient was selected for the analysis of events during the 90 days post discharge. In this study, reported excess LOS due to readmission included stay in any facility such as intensive care unit (ICU) or other ward; the excess LOS spent in ICUs was also estimated separately [
      Information Services Division (ISD) Scotland
      SMR01 – general/acute inpatient and day case.
      ]. For patients with multiple HAIs within a single admission, the first identified infection type was used in the analysis by HAI type. As a robustness check the analysis was repeated to include patients who did not survive in the 90 days post discharge (see Appendix Table A4).
      The regression model controlled for other covariates likely to explain variation in the outcomes. Covariates were selected by reviewing the Scottish Patients at Risk of Readmission and Admission (SPARRA) tool which predicts an individual's risk of being admitted to hospital as an emergency inpatient within the next year [
      Information and Statistics Division
      Scottish patients at risk of readmission and admission (SPARRA).
      ,
      • Wammes J.J.G.
      • van der Wees P.J.
      • Tanke M.A.C.
      • Westert G.P.
      • Jeurissen P.P.T.
      Systematic review of high-cost patients’ characteristics and healthcare utilisation.
      ]. The covariates included were: type of hospital, sex, age, specialty of admission, type of admission (elective or emergency), Scottish Index of Multiple Deprivation (SIMD) as an indicator of deprivation, time since last inpatient procedure, LOS in the two years before admission, and comorbidities: cancer, cardiovascular disease, chronic renal failure, and diabetes.

      Valuation of HAI excess post-discharge resource use

      The second step of the analysis was to build a cost model with parameters that capture events representing extra use of services. Estimates of the regression model were multiplied by the unit costs of services to calculate post-discharge costs for the average individual affected by HAI. Publicly available unit costs were used as shown in Table I for each patient outcome. Costs are reported in pounds sterling using 2018 data [
      Information Services Division (ISD) Scotland
      R040: Specialty Group Costs – Inpatients in all Specialties (exc Long Stay) April 2017 – March 2018.
      ,
      Personal Social Services Research Unit
      Unit costs of health & social care.
      ].
      Table ICosts (£) for each NHS service
      ResourceMean unit costSD: plausible rangeDistribution for sensitivity analysisSource
      Acute care
       Bed-day ward
      Unit costs are based on NHS reference (accounting) costs.
      486.40307.08Log–normalISD [
      Information Services Division (ISD) Scotland
      R040: Specialty Group Costs – Inpatients in all Specialties (exc Long Stay) April 2017 – March 2018.
      ]
       Bed-day ICU
      Unit costs are based on NHS reference (accounting) costs.
      1800.41310.48Log–normalISD [
      Information Services Division (ISD) Scotland
      R040: Specialty Group Costs – Inpatients in all Specialties (exc Long Stay) April 2017 – March 2018.
      ]
       Theatre cost per case
      Unit costs are based on NHS reference (accounting) costs.
      616.41676.68NormalISD [
      Information Services Division (ISD) Scotland
      R040: Specialty Group Costs – Inpatients in all Specialties (exc Long Stay) April 2017 – March 2018.
      ]
       Outpatient visit134.0096.00–160.00NormalPSSRU [
      Personal Social Services Research Unit
      Unit costs of health & social care.
      ]
      Community care
       GP appointment28.0022.40–33.60NormalPSSRU [
      Personal Social Services Research Unit
      Unit costs of health & social care: II. Community based health care staff.
      ]
      NHS, National Health Service; SD, standard deviation; ISD, Information Services Division (ISD) Scotland; ICU, intensive care unit; GP, general practitioner; PSSRU, Personal Social Services Research Unit.
      a Unit costs are based on NHS reference (accounting) costs.
      Bed-day unit costs were based on reference costs that included permanent staff and other direct costs excluding capital and overhead costs [
      Information Services Division (ISD) Scotland
      R040: Specialty Group Costs – Inpatients in all Specialties (exc Long Stay) April 2017 – March 2018.
      ]. The bed-day costs considered in this study are a proxy for the direct cost of providing healthcare services in the NHS. The cost of repeat surgeries is the weighted average of theatre costs across teaching and large general hospitals in NHS Scotland [
      Information Services Division (ISD) Scotland
      R040: Specialty Group Costs – Inpatients in all Specialties (exc Long Stay) April 2017 – March 2018.
      ]. Outpatient visits were costed using the Personal Social Services Research Unit (PSSRU) [
      Personal Social Services Research Unit
      Unit costs of health & social care.
      ]. The PSSRU also provides estimates for lower and upper quartiles of the outpatient visit costs and these were used as a plausible range.
      The cost of prescriptions for antibiotic treatment within the first 90 days after the index admission episode was calculated from linkage to the Prescribing Information System dataset [
      Information Services Division (ISD) Scotland
      Prescribing information system for Scotland (PRISMS).
      ]. Community prescribing has been reported in terms of cost, number of items, number of prescriptions and defined daily dose (DDD) [
      World Health Organization
      Anatomical therapeutic chemical (ATC)/defined daily dose (DDD) toolkit.
      ]. Since there is currently no routinely available data set that collects GP visits, the number of prescriptions for an antibiotic was used as a proxy for GP visits. Cost of GP appointments was based on the PSSRU with a plausible range of ±20% [
      Personal Social Services Research Unit
      Unit costs of health & social care: II. Community based health care staff.
      ].

      Estimation of post-discharge excess costs due to HAI

      The statistical analysis accounted for the uncertainty in the unit costs and excess resource use parameters. First, an empirical distribution was determined by drawing Monte Carlo samples from log–normal and normal distributions shown in Table I. Second, a non-parametric bootstrap approach was used to calculate confidence intervals (CIs) of estimated resource use. The post-discharge cost of HAI was defined as the product of vectors of unit costs and excess resource use by HAI patients. The empirical distributions were combined with the bootstrapped data to calculate 95% CIs. Finally, the cost for a case of HAI was multiplied by the annual estimated number of HAIs in NHS Scotland. Data sources and how these were combined are shown in Figure 1.
      Figure 1
      Figure 1How information was calculated and combined in order to estimate cost per case and total cost of HAI within 90 days post discharge. HAI, healthcare-associated infection; LOS, length of stay; ECONI, Evaluation of Cost of Nosocomial Infection; PSSRU, Personal Social Services Research Unit; GP, general practitioner; NHS, National Health Service.

      Results

      The final sample included a total of 44,505 patients (see Appendix Figure A1 for a strobe diagram). A slightly higher proportion of HAI patients (8% of HAI patients) died in the 90-daypost-discharge period than non-HAI patients (5% of non-HAI patients).
      The descriptive characteristics of the HAI and non-HAI patients in the sample are presented in Table II. HAI patients tended to be older, but the two groups had similar gender ratios. There was a greater proportion of emergency admissions compared with elective admissions within the HAI group. The HAI group showed a greater proportion of comorbidities than the non-HAI group, had more recent inpatient procedures, and a greater number of days in hospital within the two years preceding their index admission.
      Table IIDescriptive characteristics of sample
      VariableHAI patientsNon-HAI patients
      (N = 664)(N = 43,841)
      Age (years)
       <4049(7.4%)7199(16.4%)
       40–4955(8.3%)4493(10.3%)
       50–59114(17.2%)7082(16.2%)
       60–69111(16.7%)8022(18.3%)
       70–79165(24.9%)8808(20.1%)
       >80170(25.6%)8237(18.8%)
      Sex
       Female (Yes)352(53.0%)23,338(53.2%)
      SIMD
       1 (most deprived)133(20.0%)8449(19.3%)
       2204(30.7%)11,271(25.7%)
       3118(17.8%)8253(18.8%)
       490(13.6%)7327(16.7%)
       5 (least deprived)116(17.5%)8294(18.9%)
       Unknown3(0.5%)247(0.6%)
      Hospital type
       Teaching hospital607(91.4%)30,804(70.3%)
       General hospital57(8.6%)13,037(29.7%)
      Admission type
       Emergency admission (Yes)562(84.6%)32,407(73.9%)
      Specialty
       Medical267(40.2%)22,501(51.3%)
       Surgical246(37.1%)16,681(38.1%)
       HDU59(8.9%)1128(2.6%)
       ICU81(12.2%)2059(4.7%)
       Obstetrics–gynaecology11(1.7%)1472(3.4%)
      Comorbidities
       Diabetes (Yes)101(15.2%)3850(8.8%)
       Chronic renal failure (Yes)177(26.7%)4038(9.2%)
       Cardiovascular disease (Yes)331(49.9%)15,882(36.2%)
       Cancer (Yes)66(9.9%)2092(4.8%)
      Last inpatient procedure (days to admission)
       None300(45.2%)28,533(65.1%)
       ≤30104(15.7%)1348(3.1%)
       31–9080(12.1%)2185(5.0%)
       91–18053(8.0%)2609(6.0%)
       ≥181127(19.1%)9166(20.9%)
      Total LOS (days) up to 2 years prior to admission
       0235(35.4%)27,192(62.0%)
       1–261(9.2%)4342(9.9%)
       3–765(9.8%)4710(10.7%)
       8–1470(10.5%)2710(6.2%)
       15–3086(13.0%)2384(5.4%)
       >30147(22.1%)2503(5.7%)
      HAI type
      Patients with multiple HAIs during their stay were assigned to the HAI type of the first infection.
       BSI96(14.5%)N/AN/A
       GI98(14.8%)N/AN/A
       LRI122(18.4%)N/AN/A
       PN52(7.8%)N/AN/A
       SSI107(16.1%)N/AN/A
       UTI161(24.3%)N/AN/A
       Other28(4.2%)N/AN/A
      HAI, healthcare-associated infection; SIMD, Scottish Index of Multiple Deprivation; HDU, high-dependency unit; ICU, intensive care unit; LOS, length of stay; BSI, bloodstream infection; GI, gastrointestinal infection; LRI, lower respiratory tract infection; PN, pneumonia; SSI, surgical site infection; UTI, urinary tract infection; N/A, not applicable.
      a Patients with multiple HAIs during their stay were assigned to the HAI type of the first infection.
      The results of the regression model are presented in Table III. The coefficients of excess resource use, in Table III, may be interpreted as the marginal effect of HAI on post-discharge outcomes. After controlling for covariates, HAI was associated with an increase in hospital total bed-days of 2.9 days (standard error (SE): 0.3) but not the total number of readmissions (0.003; SE: 0.04), so patients with HAI when readmitted stayed for a longer period. There was no difference in the number of surgeries for patients with HAI compared to those without HAI (0.000; SE: 0.03). HAI patients showed a small increase in the number of outpatient visits of 0.371 (SE: 0.97) within 90 days after discharge compared to those without HAI.
      Table IIIMean excess resource use per HAI patient within 90 days post-discharge from hospital
      Patient outcomes 90 days post-dischargeExcess resource use
      Coefficients in this column represent point estimates. All models adjusted for type of hospital, sex, age, specialty of admission, type of admission (elective or emergency), Scottish Index of Multiple Deprivation, time since last inpatient procedure, length of stay (LOS) in the two years before admission, diagnosis of cancer, cardiovascular disease, chronic renal failure, and diabetes.
      SE95% CI
      Bias-corrected 95% CI based on 5000 bootstrap samples.
      Acute care
       No. of readmissions to hospital0.0030.0390.000–0.078
       Length of total stay (days)
      Total LOS includes stay in any facility such as intensive care, ICU, high-dependency unit, and other wards.
      2.8510.2972.518–3.672
       Length of stay in ICU (days)
      LOS in ICU is estimated separately and should not be added to total LOS.
      0.0850.0390.033–0.218
       No. of surgeries0.0000.0290.000–0.062
       No. of outpatient visits
      An individual can have more than one outpatient attendance on the same day in the sample.
      0.3710.0970.220–0.622
      Community care
       Daily defined doses
      This is daily defined doses (DDD) prescribed in the period of analysis, which may not reflect actual DDD taken.
      6.3680.6865.496–8.640
       Gross ingredient cost (£)12.7962.20210.328–23.799
       Dispensed antibiotics0.2740.0380.224–0.388
       Prescriptions for antibiotics
      One prescription may have more than one item, and each item may have different quantities of the same drug.
      ,
      Calculated from the number of prescription dates per person in period.
      0.2690.0370.220–0.360
      HAI, healthcare-associated infection; SE, standard error; CI, confidence interval; ICU, intensive care unit.
      a Coefficients in this column represent point estimates. All models adjusted for type of hospital, sex, age, specialty of admission, type of admission (elective or emergency), Scottish Index of Multiple Deprivation, time since last inpatient procedure, length of stay (LOS) in the two years before admission, diagnosis of cancer, cardiovascular disease, chronic renal failure, and diabetes.
      b Bias-corrected 95% CI based on 5000 bootstrap samples.
      c Total LOS includes stay in any facility such as intensive care, ICU, high-dependency unit, and other wards.
      d LOS in ICU is estimated separately and should not be added to total LOS.
      e An individual can have more than one outpatient attendance on the same day in the sample.
      f This is daily defined doses (DDD) prescribed in the period of analysis, which may not reflect actual DDD taken.
      g One prescription may have more than one item, and each item may have different quantities of the same drug.
      h Calculated from the number of prescription dates per person in period.
      Antibiotic costs during the 90-day period following the index hospitalization were £12.80 (SE: 2.2) higher for HAI patients compared to non-HAI patients. HAI patients were prescribed 6.4 DDD more than non-HAI patients. The total number of prescriptions for antibiotics was 0.27 greater in the HAI group and the number of dispensed antibiotic items was 0.27 greater in the HAI group. In the HAI group 41% of patients received an antibiotic within 90 days of discharge compared with 24% of non-HAI patients. HAI resulted in an increase in resource use for each outcome except number of repeat hospitalizations and surgeries where there was no difference between HAI and non-HAI patients.
      Based on results in Table III, the total excess LOS of readmitted HAI patients was the biggest estimated coefficient (2.851) by far. Excess LOS for these readmissions by HAI type is reported in Table IV. The longest additional LOS in readmissions was seen in patients with bloodstream infection (BSI), followed by gastrointestinal infection (GI), pneumonia, and urinary tract infection (UTI). SSIs and lower respiratory tract infections had the least impact on post-discharge LOS.
      Table IVHAI coefficients for 90-day post-discharge length of stay by HAI type with bias corrected
      HAI typeLength of total stay (days)
      Point estimate was reported at a patient level and excludes all patients with incomplete follow-up. Adjusted for type of hospital, sex, age, specialty of admission, type of admission (elective or emergency), Scottish Index of Multiple Deprivation, time since last inpatient procedure, length of stay in the two years before admission, diagnosis of cancer, cardiovascular disease, chronic renal failure, and diabetes.
      95% CI
      Bias-corrected 95% CI based on 5000 bootstrap samples.
      Bloodstream infection5.444.36–7.85
      Gastrointestinal infection3.802.71–5.89
      Lower respiratory tract infection1.210.28–3.09
      Pneumonia2.761.37–6.13
      Surgical site infection1.740.73–3.90
      Urinary tract infection2.661.85–4.37
      HAI, healthcare-associated infection; CI, confidence interval.
      a Point estimate was reported at a patient level and excludes all patients with incomplete follow-up. Adjusted for type of hospital, sex, age, specialty of admission, type of admission (elective or emergency), Scottish Index of Multiple Deprivation, time since last inpatient procedure, length of stay in the two years before admission, diagnosis of cancer, cardiovascular disease, chronic renal failure, and diabetes.
      b Bias-corrected 95% CI based on 5000 bootstrap samples.
      In Table V, the excess post-discharge costs for patients with HAI compared to those without HAI are shown. The total extra cost per patient with HAI was £1,457 (95% CI: 1,004–4,244) in the 90 days post discharge. The greatest cost was associated with repeat hospital admissions.
      Table VEstimated excess post-discharge costs (£) of HAI
      Patient outcomeCost per patient95% CI
      Based upon 5000 bootstrap samples for the regression estimates and 5000 Monte Carlo simulations of the costs.
      Acute care
       Cost of repeat hospital admissions1,386.58781.83–3,754.31
       Cost of number of surgeries0.000.00–41.58
       Cost of outpatient visits49.6528.58–89.09
       Subtotal1,436.23
      Community care
       Cost of antibiotics12.8010.33–23.80
       Cost of GP appointments (prescription)7.535.89–11.26
       Subtotal20.33
      Total cost per patient1,456.561,004.23–4,243.61
      HAI, healthcare-associated infection; CI, confidence interval; GP, general practitioner.
      a Based upon 5000 bootstrap samples for the regression estimates and 5000 Monte Carlo simulations of the costs.

      Discussion

      This is the first nationally representative study in Scotland to investigate the burden of HAI in the post-discharge period from an NHS perspective based on whole-hospital incidence. The results suggest that patients who develop HAI in hospital have longer LOS if they are readmitted, and they are prescribed more antibiotics in the 90 days post-discharge period than non-HAI patients. On average HAI patients were prescribed one single additional prescription of antibiotics given that most antibiotics in the community are prescribed for between five and seven days. Overall, the results of this study suggest that 95% of the post-discharge costs associated with HAI were due to hospital readmissions. These hospital costs are mainly staff costs associated with the direct cost of patient care [
      Information Services Division (ISD) Scotland
      R040: Specialty Group Costs – Inpatients in all Specialties (exc Long Stay) April 2017 – March 2018.
      ]. Excess community care costs are ∼1.4% of the total cost associated with HAI in the post-discharge period. The rest of the post-discharge cost is associated with outpatient appointments (3.4%).
      Based on the total annual numbers of HAI in teaching and large general hospitals in NHS Scotland (7437 (95% CI: 7021–7849) estimated elsewhere), annual post-discharge costs equate to approximately £11 million (9.5m–14.4m) [
      • Stewart S.
      • Robertson C.
      • Pan J.
      • Kennedy S.
      • Dancer S.
      • Haahr L.
      • et al.
      Epidemiology of healthcare-associated infection reported from a hospital-wide incidence study: considerations for infection prevention and control planning.
      ]. During 2018–2019 the total inpatient expenditure in NHS Scotland was £4.1 billion, making the post-discharge cost of treatment ∼0.3% of the total acute inpatient budget [
      Information Services Division (ISD) Scotland
      R025: Hospital running costs by patient type – NHS board level.
      ]. The direct inpatient cost due to HAI has been estimated elsewhere to be £30.1 million (14.1m–74.4m) per year [
      • Manoukian S.
      • Stewart S.
      • Graves N.
      • Mason H.
      • Robertson C.
      • Kennedy S.
      • et al.
      Bed-days and costs associated with the inpatient burden of healthcare associated infection in the UK.
      ]. This means that on average the post-discharge period increases inpatient costs by 36%.
      This study found patterns of community resource use for patients with HAIs identified in hospital comparable to those found by Plowman et al. [
      • Plowman R.
      • Graves N.
      • Griffin M.A.S.
      • Roberts J.
      • Swan A.V.
      • Cookson B.
      • et al.
      ]. This study also confirms other previous findings that HAI burden does not stop during the inpatient period but continues after hospital discharge [
      • Marchetti A.
      • Rossiter R.
      Economic burden of healthcare-associated infection in US acute care hospitals: societal perspective.
      ]. Our results are similar to those of Chopra et al. who showed that patients with Clostridium difficile infection (CDI) have 5–6 days excess LOS when readmitted [
      • Chopra T.
      • Neelakanta A.
      • Dombecki C.
      • Awali R.A.
      • Sharma S.
      • Kaye K.S.
      • et al.
      Burden of Clostridium difficile infection on hospital readmissions and its potential impact under the Hospital Readmission Reduction Program.
      ]. This study shows that GI, which could have been caused by any pathogenic agent including CDI, increases LOS in readmissions by ∼4 days. The results of this study are also similar to those of Nelson et al. who showed that patients with HAI caused by meticillin-resistantStaphylococcus aureus have increased LOS when readmitted with additional pharmacy costs [
      • Nelson R.E.
      • Jones M.
      • Liu C.F.
      • Samore M.H.
      • Evans M.E.
      • Graves N.
      • et al.
      The impact of healthcare-associated methicillin-resistant Staphylococcus aureus infections on post-discharge healthcare costs and utilization.
      ].
      The findings of this study have similarities with inpatient costs reported elsewhere which suggest that BSIs and pneumonias are the most burdensome infection types [
      • Manoukian S.
      • Stewart S.
      • Graves N.
      • Mason H.
      • Robertson C.
      • Kennedy S.
      • et al.
      Bed-days and costs associated with the inpatient burden of healthcare associated infection in the UK.
      ]. However, other infection types seem to have relatively higher costs in the inpatient period but are less costly in the post-discharge period. For example, UTIs have been estimated to have zero excess LOS in the inpatient stay but showed 2.6 days excess LOS in the post-discharge period [
      • Stewart S.
      • Robertson C.
      • Pan J.
      • Kennedy S.
      • Haahr L.
      • Manoukian S.
      • et al.
      Impact of healthcare associated infection on length of stay.
      ]. One explanation for this result may be that some UTI patients are discharged before their infections are completely treated and display complex treatment patterns if readmitted. This could also be due to unobserved patient characteristics that increase the likelihood of developing HAI but also increase hospital stay in the post-discharge period, or there might be omitted factors that confound the estimated excess LOS due to HAI. SSI has been estimated previously to be a significant burden in the post-discharge period and has also been estimated to be the third most costly infection type in the inpatient stage [
      • Manoukian S.
      • Stewart S.
      • Graves N.
      • Mason H.
      • Robertson C.
      • Kennedy S.
      • et al.
      Bed-days and costs associated with the inpatient burden of healthcare associated infection in the UK.
      ,
      • Graves N.
      • Halton K.
      • Doidge S.
      • Clements A.
      • Lairson D.
      • Whitby M.
      Who bears the cost of healthcare-acquired surgical site infection?.
      ]. This study has shown SSI to be one of the least costly infection types in the post-discharge period. Some very costly SSIs occur after hospital discharge and these were not included in this study [
      • Woelber E.
      • Schrick E.J.
      • Gessner B.D.
      • Evans H.L.
      proportion of surgical site infections occurring after hospital discharge: a systematic review.
      ]. Another explanation may be that SSIs in the post-discharge period consume other types of services such as community nursing, which was not included.
      The limitations of this study must be noted. HAIs presenting post-discharge and exclusively treated in the community would not have been identified and would potentially be within the non-HAI group. Therefore, estimates of this study are likely to be an underestimate of the true post-discharge cost of HAI. Patient-level data on community nursing visits were not available, which means that the burden on community care may also be an underestimate. Patients who either died in hospital or in the post-discharge period were excluded. This means that patients who were seriously unwell were excluded, but they may have had a different pattern of resource use. The analysis was repeated, including some of these patients (deceased in the 90 days post discharge), and the values were found to be very similar (see Appendix Table A4). The regression model may not have controlled for every factor that is correlated with both the propensity to develop HAI and the level of resource use in the post-discharge period. If that is the case, the estimates of the regression model do not uniquely capture the marginal impact of HAI but are a composite of HAI and these uncontrolled factors.
      The unit costs in this study are accounting costs, which, under conditions of spare capacity, may not reflect opportunity costs (see Manoukian et al. for a discussion of issues around using opportunity costs [
      • Manoukian S.
      • Stewart S.
      • Graves N.
      • Mason H.
      • Robertson C.
      • Kennedy S.
      • et al.
      Bed-days and costs associated with the inpatient burden of healthcare associated infection in the UK.
      ]). In this study the unit cost of a bed-day in wards was calculated by excluding more resource-intensive facilities such as ICUs and high-dependency units. Some of the excess LOS due to HAI readmissions may have taken place in these more expensive units. The regression results suggest that ∼3% of the total excess LOS due to readmissions takes place in ICUs. This implies an extra cost of £154 on average associated with readmissions of HAI patients in ICUs, meaning that the estimated post-discharge cost of HAI was underestimated. The purpose of this study was to investigate resource use in the post-discharge period but it is important to recognize that HAI also affects quality of life. Quality of life can be measured using quality-adjusted or disability-adjusted life-years but this was out of scope for this study. This study did not examine societal costs such as informal care, loss of income due to work absence, and other personal costs. These indirect costs are not routinely collected but may be substantial for some individuals affected by HAI and are not always visible to policy-makers.
      The strengths of this study are its data sources and the methodology. This study is important because it provides information on where future IPC resources could be directed to increase efficiency within the health system. A 10% reduction in HAI incidence could result in more than 2100 bed-days being made available by limiting the length of readmission episodes. It is important to note that the costs reported in this study cannot be recovered as cash savings if HAI is prevented, but are valuations of alternative uses such as treating other patients [
      • Graves N.
      • Harbarth S.
      • Beyersmann J.
      • Barnett A.
      • Halton K.
      • Cooper B.
      Estimating the cost of health care-associated infections: mind your p’s and q’s.
      ]. These findings indicate that a reduction in HAI cases would mainly impact acute care rather than community care. However, patients with HAI had on average one additional prescription for antibiotics post discharge, and although the cost of these antibiotics and the associated GP prescribing time was small, these contribute to the overall burden of community prescribing, and therein may have secondary unmeasured impacts such as a higher risk of antimicrobial resistance.
      In conclusion, this is the first study for more than 20 years in the UK to examine the cost of HAI in the post-discharge period. It presents comprehensive data on all HAI at the facility and national level. The evidence presented is consistent with existing literature and shows that HAI increases costs and antibiotic consumption after discharge. More research is needed, especially on the post-discharge impact of HAI on patients and informal carers. Economic evaluations of IPC studies should incorporate post-discharge costs, including readmission costs, which may be a substantial proportion of total costs attributed to HAI. These findings can be used nationally and internationally to support decision-making on the impact of IPC interventions.

      Acknowledgements

      Data collection could not have been completed without the enormous effort from the ECONI research nurses: N. Williamson, H. Thain, R. Boyle, A. McAlpine, and C. Beith. Thanks are extended to all members of the infection prevention and control teams, microbiology departments and research and development teams in the study sites for their support in initiating the study. The study was supported by the clinical staff in the study hospitals, particularly the infection control teams, laboratory staff, and research and development teams.

      Author contributions

      S.M. contributed to aspects of the study design and led data collection for economic analysis, calculated the unit costs and prepared the manuscript. S.S. led the study design, wrote study protocols and ethics and Public Benefit and Privacy Panel approvals, contributed to the development of statistical analysis, the manuscript, and costed the cash treatment costs. N.G. prepared the manuscript and contributed to the study design. H.M. contributed to the study design, led on health economic aspects of the study design, and contributed to the manuscript. C.R. contributed to the concept of the study, study design, statistical analysis plan, and the manuscript. S.K. and J.P. undertook the statistical analysis and prepared results. L.H. contributed to the final manuscript. S.D. and B.C. are the Principal Investigators at the recruiting sites and contributed to the final manuscript. J.R. conceived the study, is Chief Investigator for the study, and contributed to the manuscript.

      Non-author collaborators

      The ECONI Steering Committee. M.A. represented the funder, A.L., R.D., A.M., M.S. and J.I. represented the Scottish Government HAI policy unit on the Steering Committee. E.R. and L.R. represented Infection Prevention Society (IPS). M.W., L.B., and M.R. were lay representatives on the Steering Committee and M.W. and L.B. contributed to the development of the patient-facing materials for the study. Committee: Professor J. Reilly (J.R.), Professor M. Adil (M.A.), Dr H. Mason (H.M.), Professor C. Robertson (C.R.), Professor N. Graves (N.G.), J. Ives (J.I.), M. Syme (M.S.), R. Dunk (R.D.), A. Mullings (A.M.), E. Ross (E.R.), Professor S. Dancer (S.D.), Dr B. Cook (B.C.), Professor A. Leonard (A.L.), M. Whyte (M.W.), M. Rodgers (M.R.), L. Brown (L.B.), S. Stewart (S.S.).

      Conflict of interest statement

      None declared.

      Funding sources

      This work was supported by Health Protection Scotland Programme, NHS National Services Scotland (RIE reference 14-154) from October 2015 to January 2020. The work was also pump-primed by the Scottish Healthcare Associated Infection Prevention Institute (SHAIPI), which has been set up with Scottish Government funding via the Scottish Infection Research Network.

      Appendix ASupplementary data

      The following is the Supplementary data to this article:

      References

        • World Health Organization
        Report on the burden of endemic health care-associated infection worldwide.
        WHO, Geneva2011 (Available at:) ([last accessed March 2020])
        • Stone P.W.
        • Braccia D.
        • Larson E.
        Systematic review of economic analyses of health care-associated infections.
        Am J Infect Control. 2005; 33: 501-509https://doi.org/10.1016/j.ajic.2005.04.246
        • Graves N.
        • Halton K.
        • Curtis M.
        • Doidge S.
        • Lairson D.
        • McLaws M.
        • et al.
        Costs of surgical site infections that appear after hospital discharge.
        Emerg Infect Dis. 2006; 12 (05-1321 [pii]): 831
        • Perencevich E.N.
        • Sands K.E.
        • Cosgrove S.E.
        • Guadagnoli E.
        • Meara E.
        • Platt R.
        Health and economic impact of surgical site infections diagnosed after hospital discharge.
        Emerg Infect Dis. 2003; 9: 196-203https://doi.org/10.3201/eid0902.020232
        • Limon E.
        • Shaw E.
        • Badia J.M.
        • Piriz M.
        • Escofet R.
        • Gudiol F.
        • et al.
        Post-discharge surgical site infections after uncomplicated elective colorectal surgery: impact and risk factors. The experience of the VINCat Program.
        J Hosp Infect. 2014; 86: 127-132https://doi.org/10.1016/j.jhin.2013.11.004
        • Roberts R.R.
        • Scott 2nd, R.D.
        • Hota B.
        • Kampe L.M.
        • Abbasi F.
        • Schabowski S.
        • et al.
        Costs attributable to healthcare-acquired infection in hospitalized adults and a comparison of economic methods.
        Med Care. 2010; 48: 1026-1035https://doi.org/10.1097/MLR.0b013e3181ef60a2
        • Plowman R.
        • Graves N.
        • Griffin M.A.
        • Roberts J.A.
        • Swan A.V.
        • Cookson B.
        • et al.
        The rate and cost of hospital-acquired infections occurring in patients admitted to selected specialties of a district general hospital in England and the national burden imposed.
        J Hosp Infect. 2001; 47: 198-209https://doi.org/10.1053/jhin.2000.0881
        • Plowman R.
        • Graves N.
        • Griffin M.A.S.
        • Roberts J.
        • Swan A.V.
        • Cookson B.
        • et al.
        The socio-economic burden of hospital acquired infection. Vol. 1. Public Health Laboratory Service, London1999
        • Chopra T.
        • Neelakanta A.
        • Dombecki C.
        • Awali R.A.
        • Sharma S.
        • Kaye K.S.
        • et al.
        Burden of Clostridium difficile infection on hospital readmissions and its potential impact under the Hospital Readmission Reduction Program.
        Am J Infect Control. 2015; 43: 314-317https://doi.org/10.1016/j.ajic.2014.11.004
        • de Souza E.C.O.
        • Ferreira Filho S.P.
        • Gervatauskas K.M.
        • Fortaleza C.M.C.B.
        Post-discharge impact of healthcare-associated infections in a developing country: a cohort study.
        Infect Control Hosp Epidemiol. 2018; 39: 1274-1276https://doi.org/10.1017/ice.2018.201
        • Khong C.J.
        • Baggs J.
        • Kleinbaum D.
        • Cochran R.
        • Jernigan J.A.
        The likelihood of hospital readmission among patients with hospital-onset central line-associated bloodstream infections.
        Infect Control Hosp Epidemiol. 2015; 36: 886-892https://doi.org/10.1017/ice.2015.115
        • Stevens V.
        • Geiger K.
        • Concannon C.
        • Nelson R.E.
        • Brown J.
        • Dumyati G.
        Inpatient costs, mortality and 30-dayre-admission in patients with central-line-associated bloodstream infections.
        Clin Microbiol Infect. 2014; 20: O318-O324https://doi.org/10.1111/1469-0691.12407
        • Nelson R.E.
        • Jones M.
        • Liu C.F.
        • Samore M.H.
        • Evans M.E.
        • Graves N.
        • et al.
        The impact of healthcare-associated methicillin-resistant Staphylococcus aureus infections on post-discharge healthcare costs and utilization.
        Infect Control Hosp Epidemiol. 2015; 36: 534-542https://doi.org/10.1017/ice.2015.22
        • Marchetti A.
        • Rossiter R.
        Economic burden of healthcare-associated infection in US acute care hospitals: societal perspective.
        J Med Econ. 2013; 16 ([doi]): 1399https://doi.org/10.3111/13696998.2013.842922
        • Graves N.
        • Harbarth S.
        • Beyersmann J.
        • Barnett A.
        • Halton K.
        • Cooper B.
        Estimating the cost of health care-associated infections: mind your p’s and q’s.
        Clin Infect Dis. 2010; 50 ([doi]): 1017https://doi.org/10.1086/651110
        • Health Protection Scotland
        Scottish national point prevalence survey of healthcare associated infection and antimicrobial prescribing 2016.
        Health Protection Scotland, Glasgow2017 (Available at:) ([last accessed February 2020])
        • Stewart S.
        • Robertson C.
        • Manoukian S.
        • Haahr L.
        • Mason H.
        • McFarland A.
        • et al.
        How do we evaluate the cost of nosocomial infection? The ECONI protocol: an incidence study with nested case–control evaluating cost and quality of life.
        BMJ Open. 2019; 9e026687https://doi.org/10.1136/bmjopen-2018-026687
        • European Centre for Disease Prevention and Control
        Point prevalence survey of healthcare-associated infections and antimicrobial use in European acute care hospitals – protocol version 5.3.
        ECDC, Stockholm2016 (Available at:) ([last accessed June 2020])
        • Stewart S.
        • Robertson C.
        • Pan J.
        • Kennedy S.
        • Dancer S.
        • Haahr L.
        • et al.
        Epidemiology of healthcare-associated infection reported from a hospital-wide incidence study: considerations for infection prevention and control planning.
        J Hosp Infect. 2021; 114: 10-22
        • Health and Social Care Information Centre
        OPCS classification of interventions and procedures version 4.7 volume I – tabular list.
        Stationary Office, London2014
        • Information Services Division (ISD) Scotland
        SMR01 – general/acute inpatient and day case.
        2017 (Available at:) ([last accessed October 2017])
        • Information Services Division (ISD) Scotland
        SMR00 – outpatient attendance.
        2017 (Available at:) ([last accessed October 2017])
        • Information Services Division (ISD) Scotland
        Prescribing information system for Scotland (PRISMS).
        2016 (Available at:) ([last accessed July 2018])
        • Information and Statistics Division
        Scottish patients at risk of readmission and admission (SPARRA).
        2019 (Available at:) ([last accessed June 2020])
        • Wammes J.J.G.
        • van der Wees P.J.
        • Tanke M.A.C.
        • Westert G.P.
        • Jeurissen P.P.T.
        Systematic review of high-cost patients’ characteristics and healthcare utilisation.
        BMJ Open. 2018; 8e023113https://doi.org/10.1136/bmjopen-2018-023113
        • Information Services Division (ISD) Scotland
        R040: Specialty Group Costs – Inpatients in all Specialties (exc Long Stay) April 2017 – March 2018.
        2018 (Available at:) ([last accessed April 2019])
        • Personal Social Services Research Unit
        Unit costs of health & social care.
        Canterbury: University of Kent, 2018 (Available at:) ([last accessed March 2019])
        • World Health Organization
        Anatomical therapeutic chemical (ATC)/defined daily dose (DDD) toolkit.
        2019 (Available at:) ([last accessed March 2020])
        • Personal Social Services Research Unit
        Unit costs of health & social care: II. Community based health care staff.
        Canterbury: University of Kent, 2018 (Available at:) ([last accessed March 2019])
        • Information Services Division (ISD) Scotland
        R025: Hospital running costs by patient type – NHS board level.
        2019 (Available at:) ([last accessed January 2020])
        • Manoukian S.
        • Stewart S.
        • Graves N.
        • Mason H.
        • Robertson C.
        • Kennedy S.
        • et al.
        Bed-days and costs associated with the inpatient burden of healthcare associated infection in the UK.
        J Hosp Infect. 2021; 114: 43-50
        • Stewart S.
        • Robertson C.
        • Pan J.
        • Kennedy S.
        • Haahr L.
        • Manoukian S.
        • et al.
        Impact of healthcare associated infection on length of stay.
        J Hosp Infect. 2021; 114: 23-31
        • Graves N.
        • Halton K.
        • Doidge S.
        • Clements A.
        • Lairson D.
        • Whitby M.
        Who bears the cost of healthcare-acquired surgical site infection?.
        J Hosp Infect. 2008; 69: 274https://doi.org/10.1016/j.jhin.2008.04.022
        • Woelber E.
        • Schrick E.J.
        • Gessner B.D.
        • Evans H.L.
        proportion of surgical site infections occurring after hospital discharge: a systematic review.
        Surg Infect. 2016; 17: 510-519https://doi.org/10.1089/sur.2015.241