Length of stay in patients undergoing total knee arthroplasty

Mehran Mannani; Mehdi Motififard; Ziba Farajzadegan; Amin Nemati

doi:10.1016/j.jor.2022.05.018

View/Download PDF

Buy Reprints

PDF

Translate this page into:

32 (); 121-124

doi:

10.1016/j.jor.2022.05.018

Length of stay in patients undergoing total knee arthroplasty

Mehran Mannani^a, Mehdi Motififard^b, Ziba Farajzadegan^c, Amin Nemati^d,

a Medical Student, Student Research Committee, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran

b Professor of Orthopedic Surgery, Orthopedics Department, Kashani Hospital, Isfahan University of Medical Sciences, Isfahan, Iran

c Professor of Community Medicine, Department of Community and Family Medicine, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran

d Orthopedic Surgeon, Student Research Committee, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran

∗Corresponding author: Amin Nemati. nemati@med.mui.ac.ir

Received: 2022-3-26, Accepted: 2022-5-30, Published: 2022-6-2

Professor P K Surendran Memorial Education Foundation

Disclaimer:
This article was originally published by Reed Elsevier India Pvt. Ltd. and was migrated to Scientific Scholar after the change of Publisher.

Abstract

Minimizing costs associated with the care of patients undergoing total knee arthroplasty (TKA) can reduce the burden on health systems that regularly struggle with limited resources. Predicting and reducing TKA associated length of stay (LoS) can therefore be invaluable. This study aimed to determine the factors that impact LoS in patients undergoing TKA and propose a model design to predict LoS.

A retrospective study was performed on patients undergoing TKA in a tertiary teaching hospital. Patients who underwent TKA from March 2007 to March 2021 were included in the study. Data were extracted from available electronic and paper records. Variables evaluated included: patients’ demographic data, general admission data, laboratory data, transfusion, operation data, and preoperative comorbidities and medical history. Independent T-test, one-way ANOVA, and Pearson correlation were used for univariate data analysis. For multivariate analysis and model designing, multiple regression stepwise methods were used.

878 patients were included in this study. Mean LoS was 6.09 (SD = 1.83) with a median of 6 days. Factors found to have a significant effect on length of stay were age, revision surgery, Anesthesia type, intensive care unit admission, insurance, transfusion, preoperative hemoglobin level, and pre-operative platelet (Plt) count. Applying a multiple regression stepwise model to these variables showed that the following pre-operative factors can be predictive for LoS: revision surgery, sex, medical insurance, hemoglobin level, and Plt count.

It was deduced that sex, revision, pre-operative hemoglobin and Plt level and health insurance were the best predictors for LoS in patients undergoing TKA.

Keywords

Length of stay

Total knee arthroplasty

Predict

Model

Show Related Articles from PubMed

1 Introduction

Total knee arthroplasty (TKA) is one of the most effective procedures performed in patients with end-stage joint arthritis who have failed conservative treatment and is predominantly performed in elder patients¹^,² The number of surgeries has increased dramatically over the past decades, resulting in significant costs to health care systems.³

Given the many limitations within healthcare settings, reducing the associated costs in patients undergoing TKA is a priority. Costs associated with TKA may be related to pre-operative factors, surgery variables, and postoperative outcomes.

Length of stay (LoS) is defined as the duration of hospitalization before and after surgery, which has direct implications for costs related to patients undergoing TKA. Previous studies have shown a direct correlation between LoS and related costs, with significant cost savings associated with shorter LoS.⁴ Given that LoS has become a benchmark for a health systems’ efficiency, identification of factors that influence LoS is necessary in order to make positive improvements.

Previous studies have identified peri-operative factors such as age, diabetes, cardio-pulmonary diseases,^5–9 prosthesis type, allogeneic blood transfusion,¹⁰ and revision to be significant variables influencing LoS in these patients. However, there is a degree of heterogeneity within identified studies which may be attributed to social-cultural factors.

This study aimed to determine the factors that influence LoS and to design a model for the prediction of LoS in patients undergoing TKA.

2 Materials and methods

2.1

2.1 Study design and participants

The study was designed as a retrospective cohort. All patients who had undergone TKA at Kashani hospital, a tertiary teaching hospital, from March 2007 to March 2021 were included in the study. Exclusion criteria were incomplete or missing medical records or patients who had not consented to participate. The study was approved by Isfahan university of medical sciences' research ethics committee (Ethics number 194-1-336). All patients consented to the use of their information for research purposes. Medical records were reviewed by the study's investigators.

2.2

2.2 Variables

The main variable of interest was LoS, defined by subtracting the discharge date from the date of admission.

2.3

2.3 Other data collected included the following parameters

●Demographic data include age, gender, marital status, place of living. Place of living was categorized as urban, rural or other cities●General admission data: day of admission, day of surgery, medical insurance status●Laboratory data including preoperative hemoglobin (Hb) level, preoperative platelet (Plt) count, and preoperative serum creatinine level.●Operation data including arthroplasty side, american society of anesthesiology (ASA) score, anesthesia type (epidural, spinal or general), intensive care unit (ICU) admission after surgery, length of stay at ICU. ASA score was used to assess preoperative medical condition and comorbidities as documented by the anesthetics team. Perioperative blood transfused was also recorded.●Medical comorbidities and medications were identified from the documented past medical history.

2.4

2.4 Statistical analysis

Demographic data were summarized using descriptive statistics including frequencies as number (%) and mean (standard deviation). Comparison of LoS among groups was done using independent t-test and ANOVA on as needed basis. Post-hoc test was performed using the Scheffé method as appropriate.

Pearson correlation test was used to compare quantitative variables including pre-operative Hb and Plt and serum creatinine level with LoS.

Pre-operative variables influencing LoS were analyzed using multiple regression, stepwise methods to draw a formula-based model. The level of significance was set to P < 0.05.

3 Result

A total of 892 patients were enrolled in our study. Fourteen patients were excluded as outliers and not included in the analysis. The mean age was 66.5 (SD = 9.4) with a range from 19 to 95. Most of the patients were female (78%) and almost all cases were married (96.9%). Mean LoS was 6.09 (SD = 1.83) with a median of 6 ranging from 2 to 15 days.

There was no significant difference in LoS for the following factors: presence of comorbid disease, being on medication before surgery, different day of admission, day of surgery, place of living, gender, marital status, side of surgery, and ASA score. (P-value>0.05).

The following factors were shown to have a statistically significant impact on LoS: age of patient, revision surgery, anesthesia type, ICU admission during hospitalization, insurance, and transfusion (p < 0.05) (Table 1).

Table 1 Relation between LOS and categorical variables.

Variable	Category	Number (%)	LoS	P.Value
Sex^a	Male	186 (21.2%)	6.24 (±1.71)	0.15
Sex^a	Female	685 (78%)	6.03 (±1.83)	0.15
Age^b	Under 50	30 (3.4%)	6.30 (±2.42)	<0.05*
	50–59	128 (14.6%)	6.13 (±1.95)
	60–69	385 (43.8%)	5.86 (±1.72)
	70–79	266 (30.3%)	6.29 (±1.80)
	80 and older	66 (7.5%)	6.33 (±1.97)
Place of living^b	Urban	625 (71.2%)	6.29 (±2.86)	0.06
	Suburban	153 (17.4%)	6.31 (±1.94)
	Another city	89 (10.1%)	6.58 (±2.34)
Marital status^a	Single	12 (1.4%)	5.83 (±1.58)	0.59
Marital status^a	Married	851 (96.9%)	6.08 (±1.83)	0.59
Revision^a	Yes	41 (4.7%)	8.5 (±2.74)	<0.001*
Revision^a	No	837 (95.3%)	5.97 (±1.69)	<0.001*
Side^b	Left	449 (51.1%)	6.22 (±2.44)	0.54
	Right	422 (48.1%)	6.42 (±2.97)
	Bilateral	2 (0.2%)	7.50 (±0.70)
Anesthesia type^b	General	71 (8.1%)	6.17 (±1.92)	<0.001 *
	Spinal	609 (69.4)	6.18 (±2.83)
	Epidural	194 (22.1%)	6.82 (±2.54)
ICU admission^a	Yes	408 (46.5%)	6.27 (±1.68)	0.003 *
ICU admission^a	No	460 (52.4%)	5.89 (±1.91)	0.003 *
Underlying^a disease	Yes	667 (76%)	6.13 (±1.84)	0.91
Underlying^a disease	No	201 (22.9%)	5.95 (1.81)	0.91
Medication^a before surgery	Yes	609 (69.4%)	6.16 (1.87)	0.33
Medication^a before surgery	No	221 (25.2%)	5.85 (±1.73)	0.33
Insurance^a	Yes	656 (74.7%)	5.97 (±1.88)	<0.001*
Insurance^a	No	220 (25.1%)	6.44 (±1.66)	<0.001*
ASA score^b	1	195 (22.2%)	5.98 (±1.76)	0.19
	2	572 (65.1%)	6.11 (±1.83)
	3	94 (10.7%)	6.18 (±2.05)
	4	8 (0.9%)	6 (±1.19)
Transfusion^a	Yes	307 (35%)	6.74 (±1.81)	<0.001 *
Transfusion^a	No	569 (64.8%)	5.72 (±1.74)	<0.001 *

a Independent T test.

b one way ANOVA.

Using the post hoc test, it was found that the significant difference observed in the type of anesthesia in ANOVA was related to the difference between spinal versus epidural anesthesia. Similarly, the observed significant difference in age groups was between groups aged 60–69 and 70–79.

Among underlying diseases, patients with a history of rheumatoid arthritis had a significantly longer LoS than those without (P-value = 0.01). Having asthma also resulted in longer LoS than other patients (7.31 ± 2.8 versus 6.05 ± 1.8, P-value = 0.01). Patients with hypothyroidism had lower LoS in comparison to other patients (5.43 ± 1.9 versus 6.13 ± 1.8, P-value = 0.01).

There also was no significant difference between LoS of patients who had any of following disease: hypertension (P-value = 0.05), ischemic heart disease (P-value = 0.17), heart failure (P-value = 0.73), diabetes mellitus (P-value = 0.30), seizure (P-value = 0.90) and bleeding disorders (P-value = 0.9).

In reviewing the medications taken by patients before surgery, patients treated with antihypertensive drugs had longer LoS than others (6.25 ± 1.8 versus 5.86 ± 1.7, P-value = 0.002). Longer LoS was also seen in patients taking clopidogrel (8.67 ± 2.8 versus 6.07 ± 1.8,P-value = 0.015). There was a significantly lower LoS in patients who were receiving levothyroxine in comparison to those who did not (5.30 ± 1.9 versus 6.13 ± 1.8,P-value = 0.003).

There was no significant difference between LoS in patients taking any of the following medications: corticosteroids (P-value = 0.21), anti-convulsant drugs (P-value = 0.17), oral anti-diabetes (P-value = 0.16), insulin (P-value = 0.7), warfarin (P-value = 0.09) and aspirin (P-value = 0.39).

Pre-operative Hb levels among male and female patients were 13.9 (±1.7) gram/dl and 12.9 (±1.3) respectively. Serum creatinine level was 1.1 g/dl (±0.3) for males and 0.9 (±0.2) gram/dl for female patients. Pre-operative Plt count was 237000 (±71000). Among quantitative variables, preoperative Hb level and Plt count had statistically significant correlation with LoS. There was no significant correlation between pre-operative serum creatinine level and LoS (Table 2).

Table 2 Bivariate Correlations between quantitative variables.


		Length of stay	Hb before surgery	Plt before surgery	Cr before surgery
Length of stay	Pearson Correlation	1
Length of stay	P-value
Pre-operative Hb	Pearson Correlation	-0.103	1
Pre-operative Hb	P-value	0.003*
Pre-operative Plt	Pearson Correlation	0.091	−0.167	1
Pre-operative Plt	P-value	0.010*	<0.0001*
Pre-operative Cr	Pearson Correlation	0.012	−0.095	−0.159	1
Pre-operative Cr	P-value	0.750	0.009*	<0.0001*

Pre-operative factors with potential predictors of LoS were used to create a model using multiple regression stepwise analysis. Variables remained in the model by their fitness (Table 3). Fig. 1 shows the model residual histogram, normal P–P plot and scatter plot.

Table 3 Coefficients variables for predicting LoS.

	Unstandardized Coefficients		Standardized Coefficients	t	p-value	95% C.I for B
	B	Std. Error	Beta	t	p-value	95% C.I for B
(Constant)	7.997	.795		10.061	P < 0.0001	(6.436–9.557)
Revision	2.813	.290	.343	9.717	P < 0.0001	(2.245–3.382)
Insurance	-.620	.152	-.142	−4.069	P < 0.0001	(-0.920 to −0.321)
Pre-operative Hb	-.101	.046	-.081	−2.196	0.028	(-0.192 to −0.011)
Sex	-.374	.162	-.084	−2.302	0.022	(-0.692 to −0.055)
Pre-operative Plt	.002	.001	.072	2.000	0.046	(0.000–0.004)

The model residual histogram, normal P–P plot and scatter plot.

Final formula has been extracted as “y = a + b1x1 + b2x2 + b3x3+ … “using stepwise model: where,” Y″ was dependent variable,” α “was intercept, “b” was slops and X was the predicting variables.By comparing the adjusted R square model was approved.

Model equation:LoS (Days) = 7.99 + (0.34 * Revision) – (0.14 * insurance) – (0.08 * Hb) – (0.08 * sex) + (0.072 * Plt)

4 Discussion

The aim of this study was to find the main variables that have an impact on LoS in patients undergoing TKA for the purpose of prediction and model design. Having health insurance, Preoperative Hb level and Plt, gender and revision surgery were the best predictors of LoS. These factors could be used to predict the LoS and assist in hospital admission planning.

Our data showed that patients who didn't have health insurance stayed longer than those who had health insurance. The rationale is that patients who did not have health insurance worried about readmission more than those who had the insurance. A number of previous studies noted this variable as a predictor of LoS. In some studies it was found to be a predictor of discharge and re-admission.¹¹^,¹²

Results from this study were in agreement with previous data showing reduced LoS in those with higher Hb levels.¹³ The study by Abdullah et al. evaluated the effect of pre-operative anemia on LoS and revealed that anemia significantly increases LoS.¹⁴ Importantly, correcting anemia prior to admission can lead to reduced LoS.

There is no consensus in including gender as a predictor of LoS. This study found men to have a longer LoS than women which is in contrast with most previous reports. Nina et al. and Inneh et al. showed that females LoS was significantly longer than men.¹⁵^,¹⁶

Few previous studies mentioned Plt count as a predictor of LoS. We found that higher Plt count results in longer LoS. Although the correlation was weak, it shows significant impact in the stepwise model. Malpani et al. showed that both high and low Plt count increase the rate of adverse outcome in patients undergoing TKA.¹⁷

As it is shown in Table 3, revision surgery was associated with longer LoS and identified as the most powerful predictor. In a nationwide survey, Yasunaga et al. found revision surgery led to longer LoS which was in line with our finding.¹⁸

Among underlying disease and drugs, rheumatoid arthritis, asthma, hypothyroidism, antihypertensive drugs, clopidogrel and levothyroxine showed significant correlations with LoS in the univariate analysis, however none of them emerged from the stepwise model.

Fig. 1 shows the difference between predicted LoS and actual LoS. The residual plot findings indicate the strong predictive capacity of the formula which makes it generalizable to a wider clinical setting.

Strengths of this study included the protracted length of time during which data was gathered (14 years). Data was obtained from a major teaching hospital with a focus on arthroplasty research and education, located within Isfahan‘s central metropolitan area in Iran. The number of variables evaluated in this study was greater than several comparable previous studies.

This study had a number of limitations. It was a retrospective analysis of available data. Data gathering was not prospective which could impact the completeness of medical records. Implementation of arthroplasty registries can assist with accurate capture and analysis of prospective data.

In conclusion we deduce that gender, revision, health insurance and pre-operative Hb and Plt levels were the best predictors for LoS in patients undergoing TKA. male patients, not having insurance and revision surgery resulted in longer stay. Preoperative Hb had negative correlation with Los while Plt count showed a positive correlation with LoS.

Funding

This project was performed with financial support from Isfahan university of medical sciences (Grant number: 194-1-336).

Ethical approval

This study was performed in line with the principles of the declaration of Helsinki. The protocol of this study was approved by the institutional review board at Isfahan university of medical sciences (number: 194-1-336).

Consent to participate

Informed consent was obtained from all participants included in the study.

Consent to publish

Study participants consented to use their medical records in a research project and publish the result as an article.

Author contributions

Mehran Mannani: Data Curation, Writing - Original Draft, Funding acquisition; Mehdi Motififard: Investigation, Resources, Visualization; Ziba Farajzadegan: Formal analysis, Writing - Review & Editing; Amin Nemati: Conceptualization, Methodology, Supervision.

References

Adie S., Harris I., Chuan A., Lewis P., Naylor J.M., . Selecting and optimising patients for total knee arthroplasty. Med J Aust. 2019;210:135-141.
[Google Scholar]
Davies P.S., Graham S.M., Maqungo S., Harrison W.J., . Total joint replacement in sub-Saharan Africa: a systematic review. Trop Doct. 2019;49:120-128.
[Google Scholar]
Daigle M.E., Weinstein A.M., Katz J.N., Losina E., . The cost-effectiveness of total joint arthroplasty: a systematic review of published literature. Best Pract Res Clin Rheumatol. 2012;26:649-658.
[Google Scholar]
Molloy I.B., Martin B.I., Moschetti W.E., Jevsevar D.S., . Effects of the length of stay on the cost of total knee and total hip arthroplasty from 2002 to 2013. J Bone Joint Surg Am. 2017;99:402-407.
[Google Scholar]
Cizmic Z., Feng J.E., Anoushiravani A.A., Borzio R.W., Schwarzkopf R., Slover J.D., . The risk assessment and prediction tool is less accurate in extended length of stay patients following total joint arthroplasty. J Arthroplasty. 2019;34:418-421.
[Google Scholar]
Cram P., Landon B.E., Matelski J., et al . Utilization and short-term outcomes of primary total hip and knee arthroplasty in the United States and Canada: an analysis of New York and ontario administrative data. Arthritis Rheumatol. 2018;70:547-554.
[Google Scholar]
Gkagkalis G., Pereira L.C., Fleury N., Luthi F., Lecureux E., Jolles B.M., . Are the Cumulated Ambulation Score and Risk Assessment and Prediction Tool useful for predicting discharge destination and length of stay following total knee arthroplasty? Eur J Phys Rehabil Med. 2019;55:816-823.
[Google Scholar]
Jorgensen C.C., Kehlet H., Lundbeck Foundation Centre for Fast-track H., Knee Replacement Collaborative G., . Role of patient characteristics for fast-track hip and knee arthroplasty. Br J Anaesth. 2013;110:972-980.
[Google Scholar]
Kim K.Y., Feng J.E., Anoushiravani A.A., Dranoff E., Davidovitch R.I., Schwarzkopf R., . Rapid discharge in total hip arthroplasty: utility of the outpatient Arthroplasty risk assessment tool in predicting same-day and next-day discharge. J Arthroplasty. 2018;33:2412-2416.
[Google Scholar]
Jonas S.C., Smith H.K., Blair P.S., Dacombe P., Weale A.E., . Factors influencing length of stay following primary total knee replacement in a UK specialist orthopaedic centre. Knee. 2013;20:310-315.
[Google Scholar]
Keswani A., Lovy A.J., Robinson J., Levy R., Chen D., Moucha C.S., . Risk factors predict increased length of stay and readmission rates in revision joint arthroplasty. J Arthroplasty. 2016;31:603-608.
[Google Scholar]
Sharareh B., Le N.B., Hoang M.T., Schwarzkopf R., . Factors determining discharge destination for patients undergoing total joint arthroplasty. J Arthroplasty. 2014;29:1355-1358 e1351.
[Google Scholar]
Lo C.K., Lee Q.J., Wong Y.C., . Predictive factors for length of hospital stay following primary total knee replacement in a total joint replacement centre in Hong Kong. Hong Kong Med J. 2017;23:435-440.
[Google Scholar]
Abdullah H.R., Sim Y.E., Hao Y., et al . Association between preoperative anaemia with length of hospital stay among patients undergoing primary total knee arthroplasty in Singapore: a single-centre retrospective study. BMJ Open. 2017;7
[Google Scholar]
Inneh I.A., . The combined influence of sociodemographic, preoperative comorbid and intraoperative factors on longer length of stay after elective primary total knee arthroplasty. J Arthroplasty. 2015;30:1883-1886.
[Google Scholar]
Mathijssen N.M., Verburg H., van Leeuwen C.C., Molenaar T.L., Hannink G., . Factors influencing length of hospital stay after primary total knee arthroplasty in a fast-track setting. Knee Surg Sports Traumatol Arthrosc. 2016;24:2692-2696.
[Google Scholar]
Malpani R., Haynes M.S., Clark M.G., Galivanche A.R., Bovonratwet P., Grauer J.N., . Abnormally high, as well as low, preoperative platelet counts correlate with adverse outcomes and readmissions after elective total knee arthroplasty. J Arthroplasty. 2019;34:1670-1676.
[Google Scholar]
Yasunaga H., Tsuchiya K., Matsuyama Y., Ohe K., . Analysis of factors affecting operating time, postoperative complications, and length of stay for total knee arthroplasty: nationwide web-based survey. J Orthop Sci. 2009;14:10-16.
[Google Scholar]