Article Text


Association between preoperative anaemia with length of hospital stay among patients undergoing primary total knee arthroplasty in Singapore: a single-centre retrospective study
  1. Hairil Rizal Abdullah1,
  2. Yilin Eileen Sim1,
  3. Ying Hao2,
  4. Geng Yu Lin3,
  5. Geoffrey Haw Chieh Liew1,
  6. Ecosse L Lamoureux4,
  7. Mann Hong Tan5
  1. 1 Department of Anesthesiology, Singapore General Hospital, Singapore
  2. 2 Health Services Research Unit, Singapore General Hospital, Singapore
  3. 3 Yong Loo Lin School of Medicine, National University of Singapore, Singapore
  4. 4 Office of Clinical Sciences, DukeNUS Medical School, Singapore
  5. 5 Department of Orthopaedics, Singapore General Hospital, Singapore
  1. Correspondence to Dr Hairil Rizal Abdullah; hairil.rizal.abdullah{at}


Objectives Studies in western healthcare settings suggest that preoperative anaemia is associated with poor outcomes after elective orthopaedic surgery. We investigated the prevalence of preoperative anaemia among patients with primary unilateral total knee arthroplasty (TKA) in Singapore and its association with length of hospital stay (LOS), perioperative blood transfusion and hospital readmission rates.

Methods Retrospective cohort study performed in a tertiary academic medical centre in Singapore, involving patients who underwent primary unilateral TKA between January 2013 and June 2014. Demographics, comorbidities, preoperative haemoglobin (Hb) level, LOS and 30-day readmission data were collected. Anaemia was classified according to WHO definition. Prolonged LOS was defined as more than 6 days, which corresponds to >75th centile LOS of the data.

Results We analysed 2394 patients. The prevalence of anaemia was 23.7%. 403 patients (16.8%) had mild anaemia and 164 patients (6.8%) had moderate to severe anaemia. Overall mean LOS was 5.4±4.8 days. Based on multivariate logistic regression, preoperative anaemia significantly increased LOS (mild anaemia, adjusted OR (aOR) 1.71, p<0.001; moderate/severe anaemia, aOR 2.29, p<0.001). Similar effects were seen when preoperative anaemia was defined by Hb level below 13 g/dL, regardless of gender. Transfusion proportionately increased prolonged LOS (1 unit: aOR 2.12, p=0.006; 2 or more units: aOR 6.71, p<0.001). Repeat operation during hospital stay, previous cerebrovascular accidents, general anaesthesia and age >70 years were associated with prolonged LOS. Our 30-day related readmission rate was 1.7% (42) cases.

Conclusion Anaemia is common among patients undergoing elective TKA in Singapore and is independently associated with prolonged LOS and increased perioperative blood transfusion. We suggest measures to correct anaemia prior to surgery, including the use of non-gender-based Hb cut-off for establishing diagnosis.

  • Anaemia
  • Arthroplasty, knee replacement
  • length of stay
  • perioperative blood transfusion, patient blood management

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See:

Statistics from

Strengths and limitations of this study

  • Our study provides a unique perspective from a South East Asian healthcare system on anaemia and its impact on length of stay (LOS) and hospital readmission rates after unilateral, primary total knee arthroplasty.

  • We had minimal incomplete data (2.0%) due to the rigorous mandatory preoperative assessment database.

  • We found comparable adjusted OR of prolonged LOS using the WHO gender-based definition of anaemia, which defines mild anaemia in men to be haemoglobin (Hb) concentration between 11.0 g/dL and 12.9 g/dL and in women to be between 11.0 g/dL and 11.9 g/dL, and a non-gender-based definition that defines mild anaemia as Hb 11.0–12.9 g/dL for both genders. Therefore, this lends weight to the recent proposition for a non-gender-based cut-off of <13.0 g/dL in defining preoperative anaemia.

  • Our study is one of the first to control for the day of the week the surgery was done, which was recently found to be a significant predictor for LOS in arthroplasties.

  • This is a retrospective observational study, so it cannot establish a causal relationship between preoperative anaemia and adverse outcomes.

  • The selection of the 75th centile for defining prolonged LOS can be viewed as an arbitrary cut-off, although it is also used in other literature examining LOS after elective surgery.


The prevalence of preoperative anaemia in patients undergoing non-cardiac surgery is considerably high, with rates of 30.4% and 28.7% recorded in the USA and Europe ,respectively.1 This is an important health problem as even mild anaemia is associated with poor outcomes postsurgery,1–5 including prolonged length of hospital stay (LOS) and higher readmission rates.6 7

Total knee arthroplasty (TKA) is one of the most common orthopaedic procedures worldwide.2 8 Due to the low mortality and morbidity rates associated with TKA, improvement initiatives have mainly focused on reducing LOS and improving functional recovery after surgery.3 4 LOS associated with TKA is directly correlated with the total procedure cost, mainly due to provision of inpatient services.5 6 Reductions in LOS can therefore help to reduce the economic burden of TKA.7 9 Due to the fixed bed capacity, high overhead costs and the Diagnosis-Related Group reimbursement system in most health systems, the economic revenue of these systems depends on an increase in patient throughput by decreasing LOS. Additionally, unexpected prolonged LOS could negatively impact the patient’s perspective of a good surgical outcome. Hence, it is important for health systems and physicians to identify modifiable risk factors that are associated with the improved recovery and discharge of these patients. Most studies examining the association between hospital LOS after joint arthroplasties and preoperative anaemia10–17 were conducted within the western healthcare settings, with different demographics, potentially diverse discharge and rehabilitation policies from other regions in the world.

Therefore, in this study, we set out to investigate the relationship between preoperative anaemia and hospital LOS, perioperative blood transfusion as well as 30-day readmission rate among patients undergoing primary unilateral TKA in Singapore.


Institutional Review Board approval was obtained (Singhealth CIRB 2014/651/D) prior to the start of the study. The electronic medical records of all 2676 patients who underwent TKA between January 2013 and June 2014 in our institution were analysed. These clinical records were sourced from our institution's clinical information system (Sunrise Clinical Manager (SCM), Allscripts, Illinois, USA) and stored in our enterprise data repository and analytics system (SingHealth-IHiS Electronic Health Intelligence System), which integrates information from administration, clinical and ancillary healthcare systems. We generated a list of patients who underwent total knee replacements from January 2013 to June 2014 using specific surgical codes relevant to this surgery. Information from SCM included patient demographics, preoperative comorbidities such as smoking, haemoglobin (Hb) level, individual components of the Revised Risk Cardiac Index (RCRI),18 19 such as a history of previous cerebrovascular accidents (CVAs), ischaemic heart disease (IHD), congestive cardiac failure (CCF), diabetes mellitus (DM) on insulin and elevated preoperative creatinine level >2 mg/dL; American Society of Anesthesiologist Physical Status (ASA-PS) score20 ; details of the operation such as site, duration, type of anaesthesia and day of week the surgery was done21; and perioperative blood transfusion and repeat surgeries during hospital stay were also obtained. The LOS was calculated from the date of admission to the date of discharge from hospital to their home environment. Thirty-day readmission data after discharge was obtained from the clinical information system database, SCM. We filtered related readmission by the International Classification of Diseases (10th revision) diagnosis and further confirmed the cause of admission by looking up the patient’s electronic medical records. We defined the window for preoperative Hb levels to be taken at a maximum of 14 days and a minimum of 1 day before the surgery. We also defined perioperative blood transfusion to be within 2 weeks before up to 2 weeks after the date of surgery.

In our institution, most patients are admitted on the day of surgery and very infrequently, 1 day earlier for medical and/or social reasons. Routinely, all antiplatelet medications apart from aspirin are stopped for the recommended duration before the surgery. The use of intraoperative tranexamic acid infiltration to the knee joint, intravenous tranexamic acid and the placement of a drain into the joint after the surgery is not standardised. Use of cell salvage is rare. Postoperatively, all patients receive a standard hospital TKA protocol for postoperative care and discharge. This includes thromboembolism chemoprophylaxis with 40 mg once daily subcutaneous low molecular weight heparin (Clexane, Sanofi, Paris, France) on the first postoperative day, which is discontinued on discharge. Patients also receive routine physiotherapy starting from the first postoperative day, even if it falls on the weekend. They are deemed fit for discharge when there is an agreement between the surgeon and physiotherapist that the patient is medically stable and that their functional ability is sufficient to allow discharge to their home environment. This includes being able to climb up a few steps, transfer with the aid of a walking frame and to bend their operated knee close to 90°.

After excluding 51 patients with more than three missing variables, three patients who did not have preoperative Hb levels, 22 patients who underwent revision surgery and 206 patients who underwent bilateral surgery, we obtained 2394 patients in the final analysis (Figure 1). No sensitivity analysis accounting for the missing data was done due to the small number (2.0%).

Figure 1.

Flowchart showing derivation of study cohort

The primary outcome was prolonged LOS, defined as more than 6 days. This cut-off was selected as it represents >75th centile LOS of the whole sample. The use of 75th centile to define prolonged LOS is consistent with other studies.22 The secondary outcome was readmission for any reason within 30 days after surgery.

Statistical analysis

We used the WHO’s gender-based definition of anaemia severity.23 Thus, mild anaemia was defined as Hb concentration of 11–12.9 g/dL in males and 11–11.9 g/dL in females; moderate anaemia was defined for both genders to be Hb concentration between 8 and 10.9 g/dL and severe anaemia defined as Hb concentration <8.0 g/dL. Among the 567 patients (23.7%) with anaemia, 403 patients (16.8%) had mild preoperative anaemia, 159 patients (6.6%) had moderate anaemia and five patients had severe anaemia (0.2%). Due to the small incidence of severe anaemia in the sample, patients with moderate and severe anaemia were analysed together.

Patient demographics and clinical characteristics were compared between LOS >6 and <6 days (table 1). For continuous variables, mean and SD were presented, and Mann-Whitney U test was used to test the mean differences between the two groups. For categorical variables, χ2 test was used to compare the proportions between the two groups. We categorised continuous variable such as age according to their broad centile groups, body mass index (BMI) according to the WHO definition of obesity and operative duration according to ≤100 min or >100 min, which corresponds to >75th centile of the data. Multivariate logistic regression was performed to determine independent predictors for increased LOS (table 2) and perioperative blood transfusion (table 3) while adjusting for demographic, preoperative clinical risk factors and surgical factors. We repeated the logistic regression for LOS with non-gender based cut-offs for anaemia to see if it yielded significantly different results from the WHO definition of anaemia. No anaemia was defined as Hb ≥13.0 g/dL, mild anaemia as Hb 11.0–12.9 g/dL and moderate/severe anaemia as Hb <11.0 g/dL. Finally, we also constructed general linear model (GLM) with our data, taking LOS and Hb as continuous variables, to assess the effect size of every unit rise in Hb on LOS. We found little presence of collinearity between Hb level and perioperative blood transfusion and their effect on LOS based on linear regression. However, in our GLM, the addition of perioperative blood transfusion as a variable reduced the significance and effect size of preoperative Hb level, hence both models with and without perioperative blood transfusion were performed and presented. All analyses were performed using IBM SPSS Statistics V.21.

Table 1

Demographics of patients with normal LOS (≤6 days) versus those with prolonged LOS (>6 days) after primary unilateral TKR

Table 2

Variables that predict increased LOS in hospital after primary unilateral total knee replacement, based on univariate and multivariate analysis

Table 3

Factors associated with perioperative blood transfusion after primary unilateral TKR



Of the 2394 patients who underwent primary TKA during the study period, 23.7% were anaemic. Four hundred and three patients (16.8%) had mild preoperative anaemia, 159 patients (6.6%) had moderate anaemia and 5 patients had severe anaemia (0.2%). In table 1, patients with prolonged LOS had higher mean age (69.2 years) compared with patients with normal LOS (65.9 years), p<0.001. There was no significant difference in BMI or gender distribution between the two groups. However, those with prolonged LOS tended to have a higher ASA-PS score (p<0.001), higher incidence of anaemia (p<0.001), previous CVA (p<0.001), undergo repeat operation within their hospital stay (p<0.001) and received general anaesthesia compared with regional anaesthesia (p<0.01).

Length of stay

The mean LOS for all patients with TKA was 5.4 days (±4.8 days), with a median of 4 days and 75th centile of 6 days. This result is similar to previous published LOS rates of primary unilateral TKA in our institution.24

Four hundred and sixty-one patients (19.3%) had LOS more than 6 days (ie, prolonged LOS). Based on the multivariate analysis presented in table 2, the variables associated with independent elevated risk of prolonged LOS include older age >70 years (aOR 1.94, p<0.001), repeat surgery within LOS (aOR 15.22, p=0.03), having mild anaemia (aOR 1.71, p<0.001) or moderate/severe anaemia (aOR 2.29, p<0.001) and previous CVA (aOR 2.89, p=0.001). Perioperative blood transfusion of 1 unit independently increased risk of prolonged LOS by 2.12 (p=0.006), while transfusion of 2 or more units independently increased risk by 6.71 (p<0.001). Variables that reduced LOS include having regional anaesthesia (aOR 0.74, p=0.02), having surgery on Thursday and being of Malay race. BMI, gender, presence of DM on insulin, previous diagnosis of IHD, previous history of congestive heart failure or smoking and elevated ASA-PS scores were not associated with higher odds of prolonged LOS.

We repeated multivariate logistic regression for LOS with the non-gender-based Hb cut-offs as described earlier. Compared with no anaemia, mild anaemia (Hb 11.0–12.9 g/dL) had an aOR 1.39 (1.09, 1.76, p=0.007), while moderate/severe anaemia (Hb <11.0 g/dL) had an aOR of 2.35 (1.56, 3.54, p<0.001) of prolonged LOS (>6 days). These results were comparable to the findings generated with WHO definition of anaemia.

Perioperative blood transfusion

Our overall transfusion rate was 5.0%. Factors that increased risk of perioperative blood transfusion (table 3) included repeat operation during hospital stay (aOR 16.51, p=0.006), presence of mild anaemia (aOR 4.13, p<0.001) and moderate/severe anaemia (aOR 9.13, p<0.001), ASA-PS score of 3 (aOR 4.00, p=0.04) and history of CCF (aOR 7.71, p=0.001). Regional anaesthesia was found to reduce risk of perioperative blood transfusion (aOR 0.57, p=0.009).

Effect size of Hb on LOS

As shown in table 4, when perioperative transfusion was excluded, every 1 g increase in preoperative Hb reduced LOS by 0.2 days (95% CI 0.08 to 0.34, p=0.002). However, after including perioperative transfusion, every 1 g increase in preoperative Hb reduced LOS minimally by 0.07 days (95% CI −0.20 to 0.06, p=0.28).

Table 4

Effect size of every 1 g increase in Hb, with and without factoring effect of perioperative blood transfusion based on general linear model

Hospital readmission within 30 days

We had 42 cases of related readmission within 30 days of discharge out of 2394 cases (1.7%) (table 5). Due to the low rates of readmission, no further statistical analysis was done.

Table 5

Causes of related admission within 30 days of discharge after primary unilateral total knee replacement


In this retrospective cohort study of 2394 consecutive patients who underwent elective unilateral primary TKA in our centre, the prevalence of WHO-defined preoperative anaemia was 23.7%. Patients with mild preoperative anaemia had an adjusted OR (aOR) of prolonged LOS (>6 days) of 1.71, while patients with moderate/severe anaemia had an aOR of 2.29. Our findings are consistent with previous studies that show that preoperative anaemia independently increased LOS in ‘fast-track’ knee arthroplasty11 and in the ‘traditional pathway’ elective primary knee arthroplasty.25 As these studies were done in the western world health systems and variations in practice and rehabilitation protocols between countries may lead to a wide disparity in postoperative hospital stays after TKA,26–29 our study provided a perspective from a South East Asian healthcare system. In countries with advanced healthcare systems and an ageing population, the exponentially increasing number of TKAs performed yearly2 importunes that more effort is needed to identify modifiable risk factors for delayed hospital discharges.

Our study showed that every 1 g increase in preoperative Hb reduced the patient’s LOS in hospital by 0.2 days. While this number is small, the presence of preoperative anaemia is one of the strongest independent predictor of perioperative blood transfusion (mild anaemia is associated with aOR of blood transfusion of 4.13, p<0.001; moderate severe anaemia with aOR of transfusion of 9.13, p<0.001), which has a strong independent influence on prolonging LOS in our study. The transfusion of 1 unit of blood is associated with aOR of prolonged LOS of 2.12, p=0.006, while transfusion of 2 or more units has an aOR of 6.71, p<0.001. Other studies have also found preoperative anaemia to be a strong predictor for perioperative allogeneic blood transfusion in knee surgeries,30 31 which is associated with poorer outcomes and prolonged LOS.11 32 Thus, optimising preoperative anaemia becomes an important healthcare cost containment strategy, since TKAs are elective procedures that can be postponed to allow optimisation of risk factors for suboptimal outcomes.

Our study adds to the growing body of publications from Asian countries on the negative impact of preoperative anaemia on postoperative outcomes in orthopaedic and non-orthopaedic surgeries.33 34 International guidelines now recommend patient blood management programmes with early detection of preoperative anaemia to identify the cause and treat any underlying reversible causes, such as iron-deficiency anaemia.35 Such intervention can reduce postoperative blood transfusion, LOS and readmission. In particular, iron deficiency anaemia is amenable to administration of intravenous iron treatment. Froessler et al demonstrated that preoperative intravenous iron treatment of iron deficiency anaemia among patients who underwent major abdominal surgery resulted in reduction of median LOS by 3 days.36 In the UK, preoperative iron supplementation prior to elective hip/knee arthroplasty has reduced transfusion rates, 90-day readmission rates and halved median hospital LOS.37 In Australia, transfusion rates have also been reduced since similar interventions for the same surgeries.38 Unfortunately in our institution, further evaluation and management of anaemia is often left to the ordering physician’s discretion, and a more systematic approach to the detection and management of preoperative anaemia is timely.

We found comparable aOR of prolonged LOS using the WHO gender-based definition of anaemia, which defines mild anaemia in men to be between 11.0 and 12.9 g/dL and that in women to be between 11.0 and 11.9 g/dL, and a non-gender-based definition, which defines mild anaemia as Hb 11.0–12.9 g/dL for both genders. Therefore, this lends weight to the recent proposition for a non-gender based cut-off of <13.0 g/dL in defining preoperative anaemia for the purpose of patient blood management initiatives.39 40

One of the aims of our study was to identify the risk factors for 30-day hospital readmission in our study population. However, we were unable to do so due to the small incidence of 42 readmissions out of 2349 cases, which precludes any meaningful analysis. Our 30-day related readmission rates are similar to previous published rate in our institution, which is between 1.7% and 2.9% for primary unilateral TKR.24 This is lower than the 30-day all-cause readmission rates in literature of 4.0%–5.5%,41–43 because it only looks at readmission that is related to complications attributable to the surgery. Nevertheless, our readmission rates are similar to Schairer et al’s 30-day surgical readmission rate of 2.4%, of which cellulitis and surgical site infection are the most common causes. In our study, surgical site infection was also the most common cause of readmission (31.1%).

Strengths of study

One of the strengths of our study was that it is among the first few to control for the day of the week the surgery was done, which has recently been shown to affect LOS in TKA and total hip arthroplasty.21 We found that having surgery on Thursday is significantly associated with the lowest odds of prolonged LOS. We postulate that this is because patients would have received standard physiotherapy for at least 2 days, on Friday and Saturday, and may be inclined to go home on Sunday where there is more familial support at home.

Another strength of our study is that we included both subjective measures of clinical risk estimates such as the ASA-PS score,27 43 as well as the components of the RCRI, which comprise diagnosis of specific conditions. This allowed us to find that out of the five clinical conditions within the RCRI score, only a history of previous CVA was significantly associated with increased LOS (aOR 2.89, p=0.001). Furthermore, a history of CCF was found to be associated with increased risk of perioperative blood transfusion (aOR 7.71, p=0.001).

Finally, we focused our study on patients who underwent primary TKA rather than combining hip and knee arthroplasty together, which is more commonly done in literature, as we wanted a more homogenous study population. We excluded revision and bilateral TKA as they are known in literature to be associated with increased need for blood transfusion and LOS.44 45

Our study recruitment occurred over 1.5 years. Our recruitment period is similar to another study published on this topic, which recruited over a 2-year period.11 We feel that this is an advantage as no major changes in healthcare and discharge policy occurred during this short time frame that would influence our primary outcome of LOS.

Limitations of study

We also had a modest study cohort of 2349 patients, which is comparable with other studies.11 41 While our study was performed retrospectively, it had minimal missing data (2.0%). This may be due to the fact that our clinical data, such as the preoperative variables, were collected electronically in a mandatory, prospective manner during routine preoperative anaesthesia assessment. Due to the observational nature of the study, a causal relationship between preoperative anaemia and adverse outcomes is difficult to establish. In addition, while our selection of the 75th centile for defining prolonged LOS can be viewed as an arbitrary cut-off in the absence of a universal definition of prolonged LOS, the use of 75th centile has been done in literature before for a similar study.22 We do not have data on whether drains were placed by the surgeons during the surgery; however, a recent study done in the same institution showed that while drains were associated with a larger drop in Hb level and greater total blood loss, it did not significantly impact blood transfusion rate, LOS or 30-day readmission rates.24 We also did not have data on the prevalence of the use of intraoperative tranexamic acid infiltration into the joint, administration of intravenous tranexamic acid by the anaesthesiologists, or use of leucodepleted blood products for transfusion. Non-leucodepleted blood has been postulated to reduce transfusion related immunomodulatory effects including mortality,46 although clinical studies in cardiac surgery have not proven to be conclusive so far.47 48


In conclusion, we found the prevalence of preoperative anaemia to be 23.7% among patients who underwent primary TKA in Singapore. It was an independent risk factor for prolonged length of hospital stay and perioperative blood transfusion. We suggest measures to correct anaemia prior to surgery, including the use of non-gender-based Hb cut-off for establishing diagnosis.

Supplementary Material

Supplementary file


We would like to thank Ms Serene Ong, Academic Medicine Research Institute, Duke-NUS Medical School, for her help in manuscript preparation. We would also like to thank Ms Sudha Harikrishnan, from the Department of Anaesthesiology, for her invaluable help in data extraction. This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Lastly, we would also like to thank Dr Chan Yiong Huak, Head, Biostatistics Unit, Yong Loo Lin School of Medicine, National University of Singapore, National University Health System, Singapore, for his invaluable help in data analysis and interpretation.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
View Abstract

Supplementary materials

Related Data


  • Contributors HRA contributed to the study conception and design, data acquisition, analysis and interpretation, drafting and final approval of the manuscript. YES contributed to data analysis and interpretation, drafting, revision and final approval of the manuscript. YH contributed to data analysis and interpretation, drafting of the manuscript and final approval. GYL contributed to data acquisition, analysis and interpretation, drafting of the manuscript and final approval. GHCL contributed to the study conception and design, data analysis, revision of the manuscript and final approval. ELL contributed to data analysis and critical appraisal, revision and final approval of the manuscript. MHT contributed to the study conception and design, drafting of manuscript and final approval.

  • Funding This research was funded by the hospital department funds.

  • Competing interests None declared.

  • Patient consent As this was a retrospective analysis of medical records, waiver of patient's consent was obtained from the ethics board.

  • Ethics approval Singhealth Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Full dataset used in the analysis can be downloaded from Dryad public repository at doi:10.5061/dryad.73250

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.