Hospital costs of post-operative delirium: A systematic review Hospital costs of post-operative delirium: A systematic review

Aims: In this systematic review, the primary aim is to investigate the hospital cost burden attributed to post-operative delirium (POD). A secondary aim is to examine how patient length of stay (LOS) in hospital varies across the selected studies. Background: POD is a common occurrence after major surgery and leads to serious medical complications. It is associated with increased morbidity and double the risk of mortality from surgery compared to non-delirious patients. POD increases patient LOS in hospital and increases the economic burden on patients and the health system. Design: A systematic review was conducted. Method: Published articles in English over the period 2010 to 2020 were searched using the PubMed and MEDLINE databases. The Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guidelines were followed. The study quality and risks of bias of included studies were assessed using the Newcastle–Ottawa Quality Assessment Scale (NOS). Results: A total of 2539 published records were initially screened and ultimately ten studies were found to be relevant to the review criteria. Six studies were from the United States of America (USA) and the others from South Korea, Australia, and Canada. The additional costs for patients with POD ranged from a minimum of US$1551 to a maximum of US$23 698 compared to non-delirious patients. Costs were higher in the USA than other countries. Studies reported most surgical patients experiencing POD were aged 70 years or older which dramatically increases the risk of its occurrence and increases LOS and hospital related costs. The difference in LOS between POD and non-delirious patients ranged from 0.8 to 7.3 days and this increased significantly if POD patients were in intensive care. Conclusions: Increased LOS and increased hospital costs are strongly associated with POD after major surgery.


Introduction
Among post-operative medical complications, delirium is common and characterised by cognitive dysfunction, inattention and thinking disorder. 1,2 Delirium has two states -hyperactive and hypoactive. 3 Post-operative delirium (POD) is significantly associated with higher risk of morbidity and mortality, inferior functional recovery and extended immobilisation. 3, 4 The major factors in developing POD are advanced age, previous history of mental dysfunction, multiple medical comorbidities, acute injuries and pain. 1,[5][6][7][8] Recent reviews of its incidence reveal a wide range from 3.3 to 77 per cent among surgical and intensive care unit (ICU) patients. [9][10][11][12] Studies report that POD also leads to prolonged length of stay (LOS) in hospital and ICU, and associated increased cost of health care treatment both in hospital and after discharge. [13][14][15][16][17][18][19] The overall additional estimated cost for delirium was reported as ranging from US$806 to US$24 509 in 2019. 20 In 2021, a study in the USA reported the health care costs attributed to POD after major elective surgery for delirious patients in one year had a mean of US$146 358 (SD: US$140 469) which is significantly higher than US$94 609 (SD: US$80 648) for nondelirious patients. The annual national health care costs in the USA due to POD were estimated at US$32.9 billion (CI 95%: US$25.7 billion-US$42.2 billion). 21 An Australian study described that the cost index of hospital episodes for post-operative delirious patients was 51 per cent higher than the non-delirious patients. Post-operative delirious patients also had a higher 28-day rehospitalisation rate than their counterparts. 22 Total cost due to delirium was about AU$8.8 billion in 2016-2017 and this severe neuropsychiatric syndrome causes about 10.6 per cent of cognitive impairment (i.e. dementia) in Australia. 23 POD also increases LOS in hospital and ICU and can lead to other postoperative complications. Increased LOS in hospital and ICU attributed to POD after major surgery is significantly higher than for nondelirious patients. [24][25][26][27] Further, hospital readmission after initial discharge was also higher among patients with cognitive impairments like POD. 28 As the prevalence of POD in ICU is upwards of 80 per cent, an investigation of the cost of POD and the resultant extended LOS is needed. 29 Research evidence shows that POD is a potentially preventable medical condition. [30][31][32] The occurrence of delirium could be avoided for 30 to 40 per cent of medical emergency patients. 33 Considering the severe impact on patient's morbidity and mortality, the prevention of POD is essential to minimise the risks to the individual surgical patient and to mitigate the economic burden on the patient, health system and society. 34,35 Aims The primary aim of this study was to systematically review the literature on the hospital costs of POD over the period 2010 to 2020. A secondary aim was to examine how patient LOS in hospital varies across the selected studies.

Review design
This review involved a systematic search of studies in the PubMed, PubMed Central and Medline databases and followed the standard Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. 36 All published research articles related to delirium and post-operative delirium (POD) including reviews and meta-analysis were taken into consideration based on MeSH terms and keywords related to cost and hospital stay.

Search strategy
Of the published journal articles from 2010 to end of 2020, articles were only included if they were peer-reviewed research articles, available as fulltext, written in English and reported on one or more of the following: the post-operative delirium condition, any associated direct or indirect hospital costs, the length of stay in hospital or ICU.
The systematic literature searching occurred in two electronic databases of PubMed, and PubMed Central, and Medline. MeSH terms, key words and subject headings were used which are conceptually synonyms of delirium, POD and the direct or indirect hospitalisation cost. The OR/ AND operator was used to create the combination of searching key words. The following MeSH terms and key words with a combination of delirium and POD were used to search the literature: "economics"; "health care economics and organizations"; "cost of illness"; "cost evaluation"; "costbenefit analysis"; "health care costs"; "cost Analysis"; "cost effectiveness"; "statistics and numerical data"; "economic outcome"; "economic impact"; "medical expenditure"; "cost utility"; "costs and cost analysis"; "hospital costs"; "medical care cost"; "delirium/statistics and numerical data"; "emergence delirium/statistics and numerical data"; "care, postoperative"; "length of stay". All the outcomes were recorded and assessed through the various filtration steps according to PRISMA guidelines and the final articles were selected.

Eligibility/inclusion criteria
The preliminary outcomes of interest were increased LOS in hospital and ICU due to POD and the additional costs of hospitalisation related to POD.
Studies that did not satisfy the inclusion criteria were excluded. Moreover, published articles not in English, systematic reviews, metaanalyses, editorials, conference proceedings, commentaries and research protocols related to delirium were also excluded.

Screening process
For this systematic review, the direct and indirect cost data and the LOS information of delirious patients were collected from selected fulllength research articles written in English. To perform this, the outcome records from the database search were evaluated by two independent researchers screening the title, abstract and the full-length articles to select the most relevant studies. This was done using the PRISMA guidelines. The first researcher (MPM) did the primary extraction and selection and discussed these with KA and JG to resolve if any conflict arose. The final selected papers were shared and evaluated by KA and JG independently. One study was excluded at the last stage due to disagreement among the researchers. This procedure ensures selection reliability and reduces the risk of bias. For each of the ten finally selected articles, the authors, publication year, types of surgery, data collection period, places/settings/country, all hospitalisation costs, LOS, and currency were extracted.

Quality appraisal
To ensure the quality of the selected studies and reduce the risk of bias, the Newcastle-Ottawa Quality Assessment Scale (NOS) 37 was used to assess the studies. The NOS is a well-established tool for cohort study evaluation in systematic reviews and meta-analyses. The NOS not only checks the study quality (i.e. study selection and comparability between the populations) but also measures the risk of bias in study outcomes or exposure variables. A score-based evaluation, with maximum NOS score of 9, was used to assess the risk of bias and indicate the study quality with 7 or higher indicating high quality, 5 or 6 indicating fair or moderate quality and less than 5 indicating high risk of bias.

Cost values and currency conversion
The extracted cost information from the articles were in different currency values and over various time periods. To make an easy, presentable and scientific comparison, the cost data was converted by using a wellestablished conversion method, namely, purchasing power parity (PPP), using US dollars in 2020 as the conversion year for comparison purposes. 38

Literature search outcomes
The search results were collected from the electronic databases using MeSH terms and POD-related keywords. The comprehensive literature search revealed a total of 2539 published records over the period from 2010 to 2020. The final selection strategies of the eligible studies are described in Figure 1 using the PRISMA framework. After excluding duplicates and articles with missing or non-English abstracts, 1569 studies continued to the next investigation step. Subsequently, these articles' titles and abstracts were screened considering the inclusion criteria and 1510 articles were excluded. Only 59 abstracts were found to fully or partly meet the inclusion criteria and the full texts of those articles were further assessed. Eleven articles were found to satisfy the inclusion criteria with one article excluded from the analysis after discussion with all researchers. Finally, ten full-text articles met the criteria and were selected for this review (see Figure 1).

Characteristics of identified studies
All ten studies included cost information and the length of hospital stay for major surgery patients.  Table 1).
Most of the selected studies were retrospective studies. They reported upon distinct types of major surgeries while one study 22 did not declare directly any particular surgery type. The studies only considered the medical or surgical acute inpatient, not their further treatment (if any) after discharge.
For the majority of post-operative patients, delirium was assessed by well-established methods, notably, confusion assessment method (CAM), confusion assessment method for the ICU (CAM-ICU), International Classification of Diseases (9 th revision) Clinical modification (ICD-9-CM) codes and International Classification of Diseases and Related Health Problems (10 th revision) Australian modification (ICD-10-AM) codes.
The NOS scores for the selected studies show minimal risk of bias and all but one study 39 had a score of seven or higher which indicates high quality (see supplementary material).
The age distribution of POD patients for the various major surgeries indicates that they were mostly elderly people of over 50 years. The mean age of POD and non-delirious patients varied from 49 to 87 years and 36 to 87 years, respectively. In two studies, the age distribution showed that POD also developed among young people under 40 years of age. 4,40 The gender ratio of POD patients in seven studies showed that males made up more than 50 per cent of patients. Overall, the proportion of males experiencing POD ranged from 29 to 84 per cent. A significant number of women had POD after the fragility hip fracture operation (82%) and lumbar fusion (LF) or lumbar decompression (LD) operations (55.5%). 24,41 Length of stay The LOS after major surgeries was represented in two ways, namely, hospital stay and ICU stay (see Table 2). Seven studies reported inpatient LOS for hospital only, one study reported LOS for ICU only and two studies reported LOS for both hospital and ICU.
The LOS in most of the studies was represented using the mean and median along with variance/spread measurements, notably, interquartile range (IQR), standard deviation (SD) and range. Two studies reported only the mean LOS 41 and frequency distribution of LOS 40 without any other dispersion/variance measurements.
The day difference of LOS in hospital between POD and non-delirious patients ranged from to 0.8 to 7.3 days (see Figure 2). The maximum mean LOS in hospital was found to be 20.2 days (SD ±13.6 days) for osteoporotic hip fractures surgeries for POD patients. 44

Costs due to POD
Eight studies used the mean 4,24,39-41, [43][44][45] and two studies showed median costs. 22,42 Studies also reported 95 per cent confidence interval (CI), IQR and SD. One study reported the standard error with the mean cost 40 . Interestingly, three studies did not report any variance measurement and only reported mean cost. 4,41,43 Costs associated with POD after major surgeries and severe medical conditions were reported in several ways, notably, total or overall cost, hospitalisation and hospitalisation admission cost, index hospitalisation and admission cost and care cost (see Table 3).
There was a significant heterogeneity among the cost reporting for POD. Six studies reported 'total' or 'overall' cost [39][40][41][42][43][44] which indicated the total cost of hospitalisation without any breakup into direct or indirect treatment costs. Four studies also reported hospitalisation or hospital admission costs. 4 One study examined hip fracture surgeries 24 and reported mean care cost for POD patients as US$24 416 (IQR US$8141 -US$10 945). Another study 39 reported costs for POD as total cost and its componentspharmacy; laboratory; diagnostic radiology; respiratory, physical therapy and occupational therapy; central supply; professional, bed expenses and dialysis. That study reported that the total 30-days cumulative incremental cost due to POD was US$20 105 (95% CI US$12 547 -US$26 484) and the incremental cost effect of mortality was US$5245 (95% CI US$2317 -US$8869) for surgical or medical ICU patients suffering from respiratory failure or shock. 39 The cost differences between POD and non-delirious patients ranged from US$1551 to US$23 698 (see Figure 3) for osteoporotic hip fracture surgery 44 and transcatheter and surgical aortic valve replacement surgeries, 45 respectively.

Discussion
In this systematic review a total of ten studies that met the inclusion criteria were reviewed. These studies had information about the extra LOS in hospital and ICU after major surgery and the associated hospital costs for an episode of POD. The studies reported the incidence of POD varied widely from 0.8 to 78.5 per cent which, in part, is explained by different study settings, study population characteristics, types of surgeries as well as the delirium diagnostic methods used after surgery. The delirium assessment method employed to identify POD might also have an impact on the extent of diagnosis of POD. The studies which used CAM as a POD diagnostic tool had greater numbers of delirious cases (16.7 to 78.5 per cent) compared to other methods like ICD-9-CM codes and ICD-10-AM codes (0.8 to 10.9 per cent). These outcomes demand a deeper investigation of POD assessment methods.
Age has been identified as a predominant factor for the occurrence of POD. 46,47 An age of 70 years or more is a well-recognised risk factor for POD which influences post-operative comorbidities and recovery. [47][48][49][50][51][52] It was observed that the older patients were the more likely they were to experience POD. Most studies reported on patient groups older than 70 years. Conversely, Patel 40 reported that a significant number of young neuro-AIDS patients (<44 years) also experienced POD (~31%). It was also observed in five studies 4,22,40,42,43 , that male patients were more affected by POD that female patients. Therefore, gender specific interventions for aged people who undergo major surgery should be undertaken to minimise the risk of POD.
All the costs reported in the studies were found to be significantly higher in POD patients compared to those who were not delirious. Kim,44 43 and after cardiac surgery was 45.9 hours longer for POD patients than non-delirious patients. 42 Four studies were conducted in the USA in same year, 2018, and reported distinct costs and LOS for different surgeries.
All studies reflected that LOS in hospital after major surgery was increased for POD patients compared to non-delirious patients.

Study limitations
First, the studies were selected from the PubMed and MEDLINE databases only. The number of studies might increase if other databases had been explored. Secondly, the timeframe for searching the studies covered only the past ten years (2010-2020) which might be a limitation to finding more studies based on the inclusion criteria. The results show that most of the studies were conducted very recently (i.e. 2017-2019) and were mainly (six out of ten) from the USA. A few studies were conducted in Asia and Australia and no studies were found from Europe and Africa. Finally, most of the studies adopted a retrospective study setup and considered the costs and LOS data from 2002 to the most recent year 2016. Furthermore, only peerreviewed and publicly available English articles were considered. This study only focused on the cost and LOS due to POD, therefore further in-depth investigation of other factors associated with POD will be informative.

Conclusions
This systemic review revealed ten studies captured the cost burden and LOS in hospital and ICU for surgical patients who developed POD. The selected studies were conducted mostly in the USA with two in South Korea and single studies in Australia and Canada. Surprisingly, no Europe studies were sighted. The present review clearly identified and summarised that hospital costs and LOS significantly increase due to POD. Although the cost increment/quantum because of POD was lower in Asia, it was extremely high in the USA studies. The highest cost due to POD was reported for the trans-catheter and surgical aortic valve replacement in USA and lowest cost in South Korea for osteoporotic hip fractures. Further clinical investigations are needed to decipher the detailed and distinct cost drivers related to POD. The present findings clearly indicate that total costs of treatment are increased with the occurrence of POD after major surgeries. This review also suggests that a gender specific investigation could be warranted as well as a deeper investigation of POD assessment methods. The outcomes of this review should be helpful for policy development regarding the different health care settings and specific cost drivers aimed at diminishing the overall costs of POD and the risk of its occurrence in surgical and hospital settings.