Advertisement

European Geriatric Medicine

, Volume 10, Issue 1, pp 67–78 | Cite as

The hospital cost of hip replacement for old inpatients in Belgium

  • Julie De FoorEmail author
  • Philippe Van Wilder
  • Pol Leclercq
  • Dimitri Martins
  • Magali Pirson
Research Paper

Abstract

Introduction

The objectives of this research are (i) to describe the medico-administrative characteristics of inpatients aged 65 and more who are hospitalized for hip joint replacement, (ii) to evaluate the complete hospital cost into costs of medical procedures, drugs costs, prostheses costs, and the administrative costs, and (iii) to identify and to evaluate from administrative database predictors influencing the complete hospital costs.

Methods

The study was based on 961 inpatient stays aged 65 and more, with the APR-DRG 301 “Hip joint replacement”. The sample for this study was based on data collected in 2014 among nine Belgian general hospitals. We used the linear regression method for isolating predictors of hospital cost.

Results

The study highlights three different types of patients hospitalized for hip replacement, depending on the primary diagnosis: osteoarthritis problems (57%), femur neck fracture (30%), or other reasons (13%) (complications, infections, or problems with the existing hip prosthesis). The median length of stay (P25–P75) was 9 days (6.29–20.91). The median cost (P25–P75) was 8,023.91 EUR (6678.32–13,670.78). The total cost was composed of the direct hospital cost (30%), the cost of medical procedures (31%), cost of drugs (4%), the cost of hip prosthesis (18%), and other costs (17%). The linear regression reveals that an extreme SOI or risk of mortality, an ICU stay, an in-hospital death, an index of Charlson comorbidities of 4 or 5, to be hospitalized for a hip replacement because of complications, infections, or problems with the existing hip prosthesis, and the length of stay, were predictors of an increase in hospital cost.

Conclusion

The cost is not increasing with the age of the patient, but mainly with the length of stay and the comorbidities linked to the age which are considered in the severity of illness and the Charlson comorbidities index. The hospital cost is higher for patients hospitalized for complications linked to an existing hip prosthesis than for a hip replacement linked to osteoarthritis problems.

Keywords

Hospital cost Hip fracture Hip osteoarthritis Hip replacement Older patient 

Introduction

Hip fractures [1] and osteoarthritis [2, 3] are common disorders among older adults (65 +) and usually lead to a hip replacement, which is considered as the most cost-effective intervention [4]. The Organization for Economic Co-operation and Development has reported an average of 166 hip replacements per 100,000 population in 2015 (OECD 34), and 255/100,000 in Belgium [4]. The OECD report shows that 10% of men and 18% of women aged over 60 years have symptomatic osteoarthritis [4]. One in four individuals is at risk of developing symptomatic hip osteoarthritis in their lifetime [5]. Concerning the fractures, almost one-third of old patients living at home will at least fall once a year [6] and 10–20% of falls cause injuries such as fractures or head traumas [1, 5]. Hip fracture or proximal femoral fracture is one of the commonest reasons for admission to an orthopedic trauma ward [7]. Almost 30% of patients admitted to hospital with a proximal femoral fracture come from care and nursing homes [7] and hospital admissions linked to a fall are often the precipitating event admissions to long-term institutional care [5].

Hip replacement mainly concerns old people as hip fracture and osteoarthritis prevalence increase with age [2, 4, 8]. The ageing of the population and the extension of the age range for these treatments [9, 10] have led to an increased number of inpatients admitted to hospital for hip replacement surgery [3, 11]. In Belgium, the number of total hip replacement has increased by 3.45% per year between 1990 and 2013 [12]. Therefore, this increase in aged-patient hospitalized for hip fracture and the great frequency of this surgery will have an impact on the related cost of care [7, 11, 13, 14]. In 2014, there were 27,609 hospital stays (all age) for a hip replacement which represented a mean cost for the Belgian social security of 9747 EUR per stay [15]. Some studies identified the determinants of costs and length of stay for hip replacement [1, 7, 9, 10, 13, 16, 17]. According to those studies, costs and LOS linked to a hip replacement or a hip fracture are lower for patients receiving surgery on the same day of admission [1], and discharged to their own home [7]. On the contrary, costs and LOS are higher for patients with specific comorbidities, and those transferred between hospitals or readmitted within 28 days [7]. The main cost drivers are the type of prosthesis [10, 12], the total number of medical procedures, and total number of different medications [13]. However, the studies largely differ in their population, in their variables, and type of care (inpatient and/or outpatient cares) included in the model, and in their cost perspective (societal perspective, hospital perspective, and patient perspective). In Belgium, little studies have been conducted on the medico-administrative variables that impact the complete hospital cost attributable to hip replacement for old inpatients (65 +) admitted as a result of a neck fracture or osteoarthritis treatment.

The objectives of this research are (i) to describe the medico-administrative characteristics of inpatients aged 65 and more who are hospitalized for hip joint replacement, (ii) to evaluate the complete hospital cost into costs of medical procedures, drugs costs, prostheses costs, and the administrative costs, and (iii) to identify and to evaluate from administrative database predictors influencing the complete hospital costs.

Methods

The sample for this study was based on data collected in 2014 among nine Belgian general hospitals by the research centre in Health Economics, Nursing and Medical Institutions Management of the School of Public Health at the Université Libre de Bruxelles in Belgium [18]. The inpatient records used in the retrospective study were fully anonymized by the hospitals and the research team does not have any access to the name, the ID number, or personal medical files of inpatients. The data set was a compilation of inpatient information from analytical cost accounting of hospitals, medical discharge summaries, and length of stays in 2014. A complete cost per hospital stay was calculated from the hospital perspective. The hospital cost took into account the direct and the indirect costs. The direct costs were the costs linked to the admission of the patients in the care unit (nursing costs, etc.). The indirect costs considered costs of medical procedures, pharmaceutical treatments, prostheses and implants, hostel costs, and administrative costs [18].

In 2014, the nine studied hospitals recorded 155,125 stays, including 42,397 inpatient stays aged 65 or more. The total number of inpatient stays of the nine studied hospitals represents 9% of total Belgian inpatient stays (all age and all diseases) [15]. The study was based on 961 inpatient stays aged 65 and more, with the APR-DRG 301 “Hip joint replacement” (All-Patient Refined Diagnosis-Related Groups, version 28).

Variables

The studied variables were issued from activity and administrative data (medical discharge summaries) of the hospitals. As the variables were predefined, we were limited and did not have the flexibility to add other variables in the data set. More precisely, the independent variables that describe medical-administrative characteristics of inpatients and hospital costs were:
  • Age is grouped in three categories: 65–74, 75–84, 85 years and over. There was no consensus in the literature on the threshold defining elderly and on age categorization among old people. The chosen classification (three groups) allowed highlighting the evolution of patient specificities with age. Inpatients aged 65 and more were called old patients and aged 85 and more the very-old inpatients.

  • Gender.

  • Severity of illness (SOI): APR-DRG is linked to a SOI subclass. Severity of illness is based on secondary diagnosis and comorbidities associated with principal diagnostic on admission [19] (www.3m.com/product/information/All-Patient-Refined-DRG-Software). There are four levels of severity: 1—minor, 2—moderate, 3—major, and 4—extreme.

  • The risk of mortality: risk of mortality is defined as the likelihood of dying [19] (www.3m.com/product/information/All-Patient-Refined-DRG-Software). There are 4 levels of risk of mortality: 1—minor, 2—moderate, 3—major, and 4—extreme.

  • The place before admission to the hospital. The patient can come to the hospital from home, from another hospital, from a nursing home, or from another place (school, work, traffic, on its way to work, public place, psychiatric home, during sporting activities, admission from a long-term stay in a merger, and others).

  • The incentive to hospitalization. The patient can go to the hospital from its own initiative, or can be sent by a general practitioner, by a specialist doctor, by a doctor on duty, or by a third party (insurance institution or outsiders).

  • Elective or emergency admission.

  • Readmission or not in the same hospital throughout the 365 last days.

  • The primary diagnosis. The patient is hospitalized for a hip joint replacement after a fracture, because of osteoarthritis problems, or for other reasons (mainly complications, infections, or problems with the existing hip prosthesis).

  • Transit or not through intensive-care unit (ICU).

  • Transit or not through geriatric unit (G).

  • Charlson comorbidity index (CCI). Hude Quan [20] developed an enhanced ICD-10 and ICD-9-CM coding algorithms to define Charlson comorbidities index [21] in medical and administrative data. We used primary and secondary diagnosis in each discharge records’ patients aged 65 and more, and we consider comorbidities present on admission and occurring during the hospitalization. We calculated the frequencies of comorbidities in the sample. A higher index shows a high number of comorbidities. The comorbidity scores were grouped in 4 categories: 0, 1–3, 4–5, 6 and over [22].

  • Destination after discharge. After hospitalization, inpatient can return home, be sent to another hospital, be admitted in a nursing home, and be admitted in a psychiatric institution or die in hospital. We isolated the variable die or not at hospital (in-hospital death).

  • The hospital length of stay (LOS) in days.

The dependent variable was the cost per stay from hospital perspective [18].

Statistics

The description of the sample was based on 961 hospital stays. To explore differences in characteristics across age categories or primary diagnosis categories, Pearson Chi-square test was used. For some variables, the Cochran–Armitage test was done to assess the presence of a linear trend between categorical variables and age categories. The relative risk ratio was presented with a confidence interval of 95%.

Then, we tested the effects of predictors on the total cost in separate univariate models. For assessing differences in median costs, the Wilcoxon test (if variable with 2 categories) or the Kruskal–Wallis test (if variable with more than two categories) followed by pairwise multiple comparisons (Bonferroni method) was applied. The Spearman’s nonparametric rank correlation coefficient revealed the degree of association between one dependent variable and another one. We finally considered the adjustment of the variables with regard to each other in a multivariate model. We used the linear regression method for isolating predictors of hospital cost. For the multivariate model, we log-transformed the cost and length of stay data to improve the normality of residuals and avoid heteroscedasticity (heteroscedasticity test of Breusch–Pagan/Cook–Weisberg). A stepwise procedure, with a probability for entering variables equal or lower than 0.05, and a probability for removal equal or greater than 0.10 were applied. Regarding the multitude of different variables, there was a risk of collinearity which was mainly detected by the stepwise procedure in STATA.

Results were presented as median with the 25th and 75th percentiles (P25–P75). Despite asymmetry of data, we sometimes also presented the mean cost with the standard deviation (SD) to compare with the existing literature. A two-sided p value of < 0.05 was considered statistically significant.

Statistical analyses were performed using the software package STATA/IC (version 13.1) and Excel (version 2013).

Results

Descriptive statistics for patient characteristics split by primary diagnosis are reported in Table 1. Cost and LOS are reported in Tables 2, 3, 4, and 5, providing median and percentiles 25 and 75 of the distribution.
Table 1

In-patient characteristics by primary diagnosis

 

Fracture

Osteoarthrosis

Others

Total

p value

N

%a

n

%a

n

%a

n

%a

Hospital stays (n = 961)

287

30

548

57

126

13

961

100

 

Age categories

        

< 0.001

 65–74

43

15

269

49

46

37

358

37

 

 75–84

115

40

233

43

63

50

411

43

 

 85+

129

45

46

8

17

13

192

20

 

Gender (n = 954)

        

0.105

 Female

206

72

350

65

82

66

638

67

 

 Male

81

28

192

35

43

34

316

33

 

Place before admission (n = 954)

        

< 0.001

 At home

195

68

536

99

105

84

836

88

 

 Other hospital

7

3

0

0

5

4

12

1

 

 Nursing home

67

23

6

1

10

8

83

9

 

 Othersb

18

6

0

0

5

4

23

2

 

Admission to hospital (n = 942)

        

< 0.001

 Elective

6

2

530

99

85

69

621

66

 

 Emergency

276

98

7

1

38

31

321

34

 

Incentive to hospitalization (n = 953)

        

< 0.001

 On its own initiative

71

25

13

2

11

9

95

10

 

 General practitioner

61

21

3

1

9

7

73

8

 

 Specialist

17

6

525

97

89

71

631

66

 

 Doctor on-call

10

4

0

0

2

2

12

1

 

 Othersc

127

44

1

0

14

11

142

15

 

Severity of illness (SOI) (n = 961)

        

< 0.001

 1: Minor

127

44

215

39

11

9

353

37

 

 2: Moderate

104

36

286

52

39

31

429

44

 

 3: Major

34

12

44

8

73

58

151

16

 

 4: Extreme

22

8

3

1

3

2

28

3

 

Risk of mortality (n = 961)

        

< 0.001

 1: Minor

96

33

423

77

67

53

586

61

 

 2: Moderate

125

44

103

19

38

30

266

28

 

 3: Major

52

18

20

4

17

14

89

9

 

 4: Extreme

14

5

2

0

4

3

20

2

 

Through intensive-care unit (n = 961)

        

< 0.001

 No

258

90

532

97

97

77

887

92

 

 Yes

29

10

16

3

29

23

74

8

 

Through a geriatric unit (n = 961)

        

< 0.001

 No

216

75

542

99

115

91

873

91

 

 Yes

71

25

6

1

11

9

88

9

 

Destination after hospitalization (n = 954)

        

< 0.001

 At home

158

55

482

89

91

73

731

77

 

 Other hospital

19

7

32

6

9

7

60

6

 

 Othersd

3

1

5

1

3

2

11

1

 

 Death

15

5

0

0

6

5

21

2

 

 Nursing home

92

32

23

4

16

13

131

14

 

Destination after hospitalization for patients coming from home (n = 836)

         

 At home

136

70

482

90

83

79

701

84

 

 Other hospital

16

8

32

6

8

8

56

7

 

 Othersd

2

1

5

1

2

2

9

1

 

 Death

11

6

0

0

4

4

15

2

 

 Nursing home

30

15

17

3

8

8

55

7

 

In-hospital death (n = 954)

        

< 0.001

 No

272

95

542

100

119

95

933

98

 

 Yes

15

5

0

0

6

5

21

2

 

Charlson Comorbidity Index (n = 961)

        

< 0.001

 0

173

60

449

82

89

71

711

74

 

 1–3

97

34

93

17

34

27

224

23

 

 4–5

14

5

4

1

1

1

19

2

 

 6+

3

1

2

0

2

1

7

1

 

aThe percentage represents the number of items/n by category (except for the hospital stays)

bOther places before admission: at school, admission from a long-term stay in a merger, at work, others, traffic, during sporting activities, psychiatric home, on its way to work, and public place

cOther incentives to hospital: insurance institution, and other persons

dOther destinations after hospitalization: others, and psychiatric home

Table 2

Median length of stay per gender, primary diagnosis, discharge destination, and age categories

 

n

Median LOS (P25–P75)

CV

p value (Kruskal–Wallis test)

All patients

961

9 (6–21)

1.23

 

Gender (n = 954)

   

0.0001

 Female

638

10 (7–25)

1.03

 

 Male

316

7 (5–16)

1.65

 

Age categories (n = 961)

   

0.0001

 65–74

358

7 (5–10)

1.63

 

 75–84

411

10 (7–24)

1.2

 

 85+

192

17 (9–31)

0.77

 

Primary diagnosis (n = 961)

   

0.0001

 Fracture

287

16 (10–33)

1.15

 

 Osteoarthritis

548

7 (5–11)

1.03

 

 Others

126

13 (7–27)

1.05

 

Through a geriatric unit (n = 961)

   

0.0001

 No

873

8 (6–16)

1.28

 

 Yes

88

30 (20–38)

0.72

 

 Destination after hospitalization (n = 954)

   

0.0001

 At home

731

8 (6–21)

1.32

 

 Other hospital

60

10 (8–15)

0.51

 

 Othersa

11

12 (8–31)

0.98

 

 Death

21

11 (6–21)

0.92

 

 Nursing home

131

13 (9–26)

0.84

 

Patients coming from home and going in a nursing home after hospitalization by primary diagnosis (n = 55)

   

0.0376

 Fracture

30

23 (13-39)

0.76

 

 Osteoarthritis

17

11 (9–20)

0.77

 

 Others

8

15 (13–22)

0.45

 

aOther destinations after hospitalization

Others Psychiatric home

Significance of p value < 0.05

Table 3

Cost components of patients hospitalized for hip joint replacement

All signs: p value < 0.05

*Kruskal–Wallis test followed by pairwise multiple comparisons (Bonferroni method). The significant pairwise comparisons (p value < 0.05) are shown with the signs

**Wilcoxon test

***Test of linear tendance

aOther incentives to hospital: insurance institution, and other persons

bOther destinations after hospitalization: others, and psychiatric home

Table 4

Multivariate cost analysis

 

Coef

SD

p value

Length of stay

0.5610

0.01

< 0.001

Type of admission

 Elective

0.0000

  

 Emergency

− 0.1221

0.03

< 0.001

Primary diagnosis

 Others

0.0000

  

 Fracture

− 0.0797

0.03

0.01

 Osteoarthritis

− 0.0733

0.02

< 0.001

Charlson Comorbidity Index

 0

0.0000

  

 1–3

0.0071

0.02

0.641

 4–5

0.1509

0.04

< 0.001

 6+

0.1075

0.07

0.11

Age categories

 65–74

0.0000

  

 75–84

− 0.0461

0.01

0.001

 85+

− 0.0656

0.02

< 0.001

Severity of illness (SOI)

 1: Minor

0.0000

  

 2: Moderate

− 0.0071

0.01

0.6

 3: Major

0.0086

0.02

0.704

 4: Extreme

0.1953

0.05

< 0.001

Risk of mortality

 1: Minor

0.0000

  

 2: Moderate

0.0273

0.02

0.094

 3: Major

0.0721

0.03

0.005

 4: Extreme

0.2414

0.05

< 0.001

Through intensive-care unit

 No

0.0000

  

 Yes

0.2468

0.02

< 0.001

Destination

 At home

0.0000

  

 Other hospital

− 0.0205

0.02

0.392

 Othersa

0.0534

0.05

0.324

 Death

0.2505

0.04

< 0.001

 Nursing home

− 0.1153

0.02

0.53

n = 941, R2 = 0.8986

aOther destinations after hospitalization: others and psychiatric home

Table 5

Multivariate LOS analysis

 

Coef

SD

p value

Place before admission

 At home

0.00

  

 Other hospital

0.42

0.23

0.07

 Othersa

− 0.04

0.14

0.08

 Nursing home

− 0.28

0.11

0.01

Age categories

 65–74

0.00

  

 75–84

0.20

0.05

< 0.001

 85+

0.26

0.07

< 0.001

Gender

 Female

0.00

  

 Male

− 0.24

0.04

< 0.001

Severity of illness (SOI)

 1: Minor

0.00

  

 2: Moderate

0.23

0.04

< 0.001

 3: Major

0.47

0.09

< 0.001

 4: Extreme

0.64

0.15

< 0.001

Risk of mortality

 1: Minor

0.00

  

 2: Moderate

0.29

0.06

< 0.001

 3: Major

0.64

0.10

< 0.001

 4: Extreme

0.38

0.18

0.035

Through intensive-care unit

 No

0.00

  

 Yes

0.31

0.10

0.002

Through geriatric unit

 No

0.00

  

 Yes

0.37

0.83

< 0.001

Destination

 At home

0.00

  

 Other hospital

0.07

0.06

0.21

 Othersb

0.09

0.15

0.566

 Death

− 0.92

0.19

< 0.001

 Nursing home

− 0.14

0.08

0.08

Incentive to hospital

 Othersc

0.00

  

 On its own initiative

− 0.21

0.1

0.034

 General practitioner

0.00

0.1

0.994

 Doctor on duty

− 0.01

0.14

0.922

 Specialist

− 0.43

0.08

< 0.001

n = 941, R2 = 0.5121

aOther places before admission: at school, admission from a long-term stay in a merger, at work, others, traffic, during sporting activities, psychiatric home, on its way to work, and public place

bOther destinations after hospitalization: others and psychiatric home

cOther incentive to hospitals: insurance institution and other persons

Patient characteristics

67% of inpatient stays hospitalized for a hip replacement were women (Table 1). The SOI was mainly minor or moderate (81% of patients). The median age (P25–P75) was 78 years (71–83).

The median length of stay (P25–P75) was 9 days (6.29–20.91) (Table 2). The study shows a statistically significant difference of the median length of stay depending on the gender, the age category, the primary diagnosis, or the destination after discharge (Table 2). For instance, we observed that the patients aged more than 85 stayed 2.4 times longer than patients aged between 65 and 74, and the inpatients going through a geriatric unit had a hospital stay 3.7 times longer than other patients. Patients discharged into a nursing home stayed 5 days longer in hospital than those discharged into their home.

Patient characteristics depending on the primary diagnosis

The study highlights that older patients (65 +) hospitalized for a hip joint replacement could be grouped into three completely different profiles depending on the primary diagnosis; osteoarthritis problems (which is the more common primary reason for a hip replacement, 57%), femur neck fracture (30%), or other reasons (13%) (mainly complications, infections, or problems with the existing hip prosthesis) (Table 1).

Patients were hospitalized, because specialists detected upstream osteoarthritis were the youngest (75% of patients aged between 65 and 74 years), they mainly came from home and went back home after hospitalization, their admission was elective (99%), the in-hospital death was 0%, and they were sent by a specialist doctor (Table 1). The median (P25–P75) LOS for patients with detected osteoarthritis was 7 days [5, 6, 7, 8, 9, 10, 11] (Table 2).

Patients admitted after a hip fracture were the oldest, were mainly admitted through the emergency (98%), stayed in a geriatric unit (25%) during hospitalization, and were mainly institutionalized after hospitalization (Table 1). Among patients admitted from home, the percentage of patients going back home after the hospitalization amounted to 70% for inpatients admitted after a fracture and 90% for inpatients admitted because of osteoarthritis problems (Table 1). The median length of hospital stay for patients hospitalized with a fracture was 16 days [10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33] (Table 2), which is significantly higher than for other patients.

Patients admitted for other reasons (mainly complications, infections, or problems with the existing hip prosthesis) were the most medically severe patients. 60% of the inpatients had a major or extreme SOI, they were mainly (67%) readmitted in the same hospital during the year, and 23% of the patients stayed in an intensive-care unit during the hospitalization (Table 1). Patients admitted for other reasons, such as complications with an existing hip prosthesis, had a hospital stay of 13 days [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27] (Table 2).

Cost analysis

The hospitalization of old inpatients (65 +) for a hip joint replacement generated a median (P25–P75) cost of 8,023.91 EUR (6,678.32–13,670.78) (Table 3). The total cost was composed of the direct hospital cost (30%), the cost of medical procedures (31%), cost of drugs (4%), the cost of hip prosthesis (18%), and other costs (17%). Correlation between age and cost was weak (Spearman’s coefficient of correlation is 0.2760, p value < 0.0001). This partially can be explained by a negative correlation between the cost of hip prosthesis and the age of the patient (Spearman’s correlation is − 0.2834, p value < 0.0001).

The univariate analyses suggest that the cost significantly differed with the categories of some variables (p value < 0.05) (Table 3). For instance, a very-old (85 +) patient would cost for hospital 1.22 times more than a patient aged between 75 and 84 and 1.44 times more than a patient aged between 65 and 74, comparing the median costs. The study also highlights a significantly different cost depending on the primary diagnosis (fracture, osteoarthritis, or other reasons). The hospitalization of old inpatients (65 +) for a hip joint replacement after a fracture generated a median (P25–P75) cost of 10,730.28 EUR (7,359.54–17,165.59), whereas it amounted to 7,341.44 EUR (6,484.46–9,105.26) for patients with osteoarthritis as primary diagnosis and to 12,834.02 EUR (7,757.42–18,186.75) for other patients (Table 3). Other factors explained the variation of hospital cost. The analyses reveal a significant association between the cost and the level of SOI, the risk of mortality, the category of Charlson Comorbidity Index (p value of the test of linear trend < 0.05), or the LOS (Spearman’s coefficient of correlation is 0.8688, p value < 0.0001). A patient hospitalized for a hip replacement associated with an extreme SOI costed 3.21 times more than a patient with a minor SOI. The study also indicates that a stay in a geriatric unit or an intensive-care unit led to a significant higher cost for hospital. Then, the patients going home after hospitalization costed significantly less than patients institutionalized after discharge (p value < 0.01). This could be due to a longer median length of hospital stay of patients institutionalized after hospitalization. The median length of stay (P25–P75) of a patient going back home was 8 days [6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21], whereas it was 13 days [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26] for institutionalized patients and 16 days [11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32] for patients coming from home and institutionalized after hospitalization.

We finally considered the simultaneous impact of the considered variables on the total cost in a multivariate model (Table 4). The linear regression (n = 941, R2 = 0.8986) reveals that, adjusted for other variables included in the model, an extreme SOI, a major or extreme risk of mortality, an ICU stay, an in-hospital death, to have an index of Charlson comorbidities of 4 or 5, to be hospitalized for a hip replacement because of complications, infections, or problems with the existing hip prosthesis (other reasons), and the length of stay, were predictors of an increase in hospital expenses for patients aged 65 and more hospitalized for a hip replacement (Table 4). On the contrary, to be readmitted in the hospital in the same year, an admission in emergency, and the older age categories were factors that decreased the hospital cost. The model automatically excluded the insignificant variables (secondary diagnosis, incentive to hospitalization, place before admission, gender, and stay in a geriatric unit).

The hospital cost was mainly influenced by the length of stay. As the correlation between LOS and cost was high, some explanatory variables could be hidden by the length of hospital stay. Thus, we made a second multivariate model to identify the predictors of length of hospital stay (Table 5). The linear regression (n = 941, R2 = 0.5121) reveals that, adjusted for other variables included in the model, a major or extreme SOI, a major or extreme risk of mortality, an ICU stay, a stay in a geriatric unit, and the older age categories, were predictors of a longer LOS for these hip replaced patients. On the contrary, to be a male, to come from a nursing home, to be sent by a specialist, or to die at hospital were factors that decreased the LOS.

Discussion

The objectives of this research were (i) to describe the characteristics of inpatients aged 65 and more who are hospitalized for hip joint replacement, (ii) to evaluate the complete hospital cost into costs of medical procedures, pharmaceutical treatments costs, prostheses costs, and the administrative costs, and (iii) to evaluate and to identify, from administrative database, predictors influencing the total hospital costs.

As hip replacement mainly concerns older people [2, 4, 8], we have focused our analysis on patients aged 65 and more. The median age of our population was 78 years. Our sample was composed of 67% of women which is comparable to the other studies [5, 13, 14, 16, 17, 23]. The demographic distribution and the fact that the risk of fall is higher for women [6] explain this overrated proportion of women. Hip replacement surgery is considered as the most common treatment for hip fractures [1] and severe hip osteoarthritis [2, 3].

Patient characteristics

Our study has highlighted three different types of patients hospitalized for hip replacement (APR-DRG 301), depending on the primary diagnosis.

Our analysis has shown a proportion of 57% of old patients (65 +) with a primary diagnosis of osteoarthritis, which seems underestimated compared to the other studies. Osteoarthritis is the most common reason for a hip replacement [4, 9] and accounts for more than 80% of all total hip replacements in Australia [5]. Those patients have registered the lowest LOS and hospital cost.

Patients admitted after a hip fracture register the longest length of hospital stay (16 days). We have estimated the median cost of a hip replacement led by a fracture to 10,730.28 EUR. The longest LOS, compared to other primary diagnosis, can be explained by two main reasons. First, our statistical analysis has revealed that a stay in a geriatric unit is a predictor of a longer LOS for patients aged 65 and more. This is mainly linked to the Belgian hospital financing system, encouraging the long length of hospital stay for geriatric patients. Second, the LOS is longer for old patients discharged into a nursing home than for patients going back home after hospitalization, as demonstrated in the other studies [7, 24]. Our study has highlighted that, among patients admitted with a hip fracture coming from home, 15% were institutionalized after hospitalization, which is line with the other articles [1, 25]. The risk of discharge into nursing home after a hip replacement increased for old patients, which has been demonstrated by another Belgian article [25]. Finally, the hip fracture is a common clinical problem and a major cause of mortality or premature death in the elderly [1, 5, 7]. Our analysis has indicated that the in-hospital mortality increased with age categories, corroborating the article of Padron-Monedero et al. [23], and reached 6% for inpatients aged 85 and more. However, some authors have pointed out that most of the deaths are not due to the hip fracture but to the associated conditions and comorbidities that affect older patients [7, 23] and that could be manifestations of fragility.

Patients admitted for other reasons (mainly complications, infections, or problems with the existing hip prosthesis) represent the highest hospital median cost, estimated to 12,834.02 EUR. Other studies have reported that hospital resource utilization for revision total hip arthroplasty is significantly higher than for primary hip replacement [8, 9, 26].

Previously, the primary diagnosis was not reflected in the Belgian reimbursement system, which was the same for all hip replacements. However, the Belgian government has set up a reform project for the financing of hospitals [27]. The reform, which should be introduced in 2019, plans to introduce a lump sum per DRG for hospital for hip replacements with an SOI 1 or 2 (medical procedures only in a first step), excluding replacements due to hip revision, infections, or fracture. These exclusions will continue to be paid with the existing financing system, a fee-for-service for the medical activity.

Cost and length of stay for a hip replacement

We have reported a median cost of 8024 EUR for a hip joint replacement surgery for old inpatients (65 +) from the hospital perspective. In 2014, hospital stays (all age) for a hip replacement represented a mean cost for the Belgian social security of 9747 EUR per stay [15]. In Sweden, the hospital mean cost of initial hip replacement, including an initial outpatient visit, the primary surgery, and the hospital stay, has been estimated to be 9740 EUR for patients aged 75 or more [16]. Stargardt [10] has compared the hospital cost for hip replacement in seven European countries (former EU-15 members) in 2005 and has calculated a mean (SD) cost of 5778 EUR (1523). However, this European study has only considered the less severe cases without any comorbidity, which are, therefore, less costly. We have reported that the total cost was mainly composed of the direct hospital cost (linked to administration) (30%), the cost of medical procedures (31%), and the cost of hip prosthesis (18%). A Belgian study of 2005 has shown that, for a hip replacement, 54% of the cost are linked to the patient admission and administration, and the hip cost represents 21% and the surgeon and anaesthetist for 11.8% [28]. Our study has shown differences of cost between hospitals, which can be explained by the difference of medical practices in Belgian hospitals [12], by the number of hospital beds which can lead to economies of scale, the number of physicians per bed and the urban location [10], the procedures that are conducted on an outpatient basis or on an in-hospital basis [10], or by the type of prosthesis and the negotiated vendor discount [17]. Some studies have shown that the cemented joint replacement (prosthesis) has the best results concerning the survival [10, 29, 30], and is the cheapest for the hospital [10], the health insurance, and the patients [12]. In Belgium, the number of uncemented hip prosthesis continuously rises, and this type of hip prosthesis accounts for 65% of the hip prosthesis in 2015 [29]. We could not analyse the type of prosthesis, but our analysis has reported a negative correlation between the cost of prosthesis and age. A Belgian study has highlighted that the number of cemented prostheses, the most expensive implants, decreases with age, while the number of hybrid prosthesis with cemented stem rises with age [30].

The multivariate analysis has highlighted that hospital cost depends on the hospital length of stay, the severity of illness, or the primary diagnosis, but decreases with the increasing age and an admission in emergency. In contrast, the univariate analysis has indicated that an older age and admission in emergency lead to a higher hospital cost. However, this is not directly linked with the age or the admission, but it is mainly due to the specific profile of the old patient; he is mainly admitted in emergency due to a hip fracture, which implies a longer length of stay, because the discharge cannot be planned and so a higher cost for hospital.

The analysis has shown a median hospital LOS of 9 days (9 hospitals). A previous Belgian study has indicated a median hospital LOS of 7 days, which varied from 5 to 11 days depending on the hospitals (2012–2013) [12]. The difference can be explained by the different scope of population included in the study (not focused on old patients like our study) and the difference in applied method. We have identified the length of hospital stay as the strongest predictor of total hospital cost, as also reported by the other studies summarized in the article of Haentjens et al. [17]. The univariate analyses have shown that the length of hospital stay depends on patient characteristics (age, extreme SOI, and extreme RM), hospital care pathway (stay in geriatric or intensive-care unit), or destination (nursing home). To reduce the cost of hip replacement, patients, hospitals, and health care actors have all their responsibilities. Hospitals have to focus on managing the initial hospital stay by, for instance, transferring more rapidly the patient to the appropriate in- or out-hospital structure (orthogeriatric care, rehabilitation center, and nursing home). A passage through an orthogeriatric unit, which proposes a multidisciplinary approach, seems to be recommended for geriatric patients with comorbidities [31] to improve health care quality [32]. The orthogeriatric model of care has proven its impact on the diminution of functional decline, on the efficacy of revalidation [33, 34], on the reduction of complications and mortality [31, 35], on the diminution of LOS [32, 36] as well as hospitalization costs [35]. Nowadays, orthogeriatric units have not been implemented in our country.

An early transfer from hospitals to another structure may ultimately be a method of cost-shifting to another institution. The cost-savings for hospitals will not necessarily reflect economies for the society [17, 28]. Finally, health care actors have to raise old patient cognitive awareness of their fragility to avoid hip fractures among old patients.

Finally, our study has encountered weaknesses. First, the variables included in the model were retrospectively extracted from hospitals database [18]. The administrative data do not allow us to collect information such as obesity, the level of physical activity, the frailty profile, the occupation of the patients [8], and more precisions about the trauma or the surgical treatment [17], which are also risk factors and predictors of cost for hospital. Those variables were not accessible as the research team has not access to the medical personal files. Neither do we have the cause of death and its relation to the hip replacement, the reason of the fracture, and the long-term outcome after hospitalization. Then, the cost calculations have to be compared carefully, because the methodology applied to calculate the cost and the perspective can differ. In our study, we have only considered the cost from hospital perspective, not from patient or social security perspective. We also do not consider outpatients’ costs (such as subsequent follow-up visits), and cost of further rehabilitation.

Conclusion

Hip replacement surgery is the last option to treat the morbidity associated with severe osteoarthritis and hip fracture. Our study links the factors affecting the cost of hip replacements for hospitals. The cost is not increasing with the age of the patient, but mainly with the length of stay and the comorbidities linked to the age which are considered in the severity of illness and the Charlson comorbidities index.

This study highlights three different profiles of patients, depending on the primary diagnosis, with a different impact on hospital cost and length of stay. The hospital cost is higher for patients hospitalized for complications linked to an existing hip prosthesis than for a hip replacement linked to osteoarthritis problems. We recommend to split the existing APR-DRG 301 into two groups: a first group for the planned hip replacement linked to osteoarthritis problems and a second group for hip revisions, infections, or fractures.

Finally, to improve health care quality and to give appropriate patient care, the role of Geriactric Day Hospital seems essential to detect geriatric patients through a comprehensive geriatric assessment (CGA) before the hospitalization. Moreover, to avoid longer LOS, Belgian hospitals have to focus on creating adapted in-hospital services such as orthogeriatric model of care, and on managing rapidly the discharge of the patient to the appropriate structure (nursing home and rehabilitation center).

Notes

Compliance with ethical standards

Conflict of interest

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Ethical approval and informed consent

The patient records used in this retrospective study were fully anonymized by the hospitals before we accessed them. The research team does not have any access to the name, the ID number, or personal medical files of inpatients. For this retrospective study, ethical approval and informed consent is not required.

References

  1. 1.
    Braithwaite RS, Col NF, Wong JB (2003) Estimating hip fracture morbidity, mortality and costs. J Am Geriatr Soc 51(3):364–370Google Scholar
  2. 2.
    Nguyen-Oghalai TU, Ottenbacher KJ, Granger CV, Smith ST, Goodwin JS (2006) Impact of osteoarthritis on rehabilitation for persons with hip fracture. Arthritis Care Res 55(6):920–924Google Scholar
  3. 3.
    Prieto-Alhambra D, Javaid MK, Judge A, Maskell J, Kiran A, de Vries F et al (2011) Fracture risk before and after total hip replacement in patients with osteoarthritis: potential benefits of bisphosphonate use. Arthritis Rheum 63(4):992–1001Google Scholar
  4. 4.
    OECD (2017) Hip and knee replacement. In: Health at a glance 2017. OECD Publishing, Paris, pp 178–179. http://dx.doi.org/10.1787/health_glance-2017-65-en
  5. 5.
    Chong EW, Wang Y, Robman LD, Aung KZ, Makeyeva GA, Giles GG et al (2015) Age related macular degeneration and total hip replacement due to osteoarthritis or fracture: melbourne collaborative cohort study. PLoS ONE 10(9):e0137322Google Scholar
  6. 6.
    Di Pierdomenico L, Uwiteka I, Senterre C, Leclercq P, da Costa EM, Pepersack T et al (2015) Évaluation économique des lésions associées aux chutes des personnes âgées vivant à domicile. Gériatrie Psychol Neuropsychiatr Vieil 13(1):36–44Google Scholar
  7. 7.
    Castelli A, Daidone S, Jacobs R, Kasteridis P, Street AD (2015) The determinants of costs and length of stay for hip fracture patients. PLoS ONE 10(7):e0133545Google Scholar
  8. 8.
    Tanna S, Wittenauer R, Smith L, Aden K (2013) Chapter 6.12 Osteoarthritis—background paper. In: Priority medicines for Europe and the world. World Health Organization. http://www.who.int/medicines/areas/priority_medicines/BP6_12Osteo.pdf?ua=1. Accessed 2 July 2018
  9. 9.
    Ackaert K, de Béthune X (2009) Prothèses totales de hanche en Belgique: analyse de suivi. Partie 1. Nombre de prothèses de hanche et coût d’une prothèses totale de hanche. MC-inform 236:24–32Google Scholar
  10. 10.
    Stargardt T (2008) Health service costs in Europe: cost and reimbursement of primary hip replacement in nine countries. Health Econ 17(S1):S9–S20Google Scholar
  11. 11.
    Moore J, Carmody O, Carey B, Harty JA, Reidy D (2017) The cost and mortality of hip fractures in centenarians. Ir J Med Sci 186:961–964Google Scholar
  12. 12.
    Anckaert K, de Béthune X (2015) MC-étude prothèse totale de hanche? Nicolas a une hanche qui flanche. MC-inform 261:23–36Google Scholar
  13. 13.
    Titler M, Dochterman J, Kim T, Kanak M, Shever L, Picone DM et al (2007) Cost of care for seniors hospitalized for hip fracture and related procedures. Nurs Outlook 55(1):5–14Google Scholar
  14. 14.
    Nikkel LE, Fox EJ, Black KP, Davis C, Andersen L, Hollenbeak CS (2012) Impact of comorbidities on hospitalization costs following hip fracture. J Bone Jt Surg-Am 94(1):9–17Google Scholar
  15. 15.
    Feedback financier par pathologie - APR-DRG 301 Remplacement d’articulation de hanche. Banque Nationale de Données Diagnostic médical/Soins & coût. (2014). https://tct.fgov.be/webetct/etct-web/html/fr/index.jsp. Accessed 7 June 2018
  16. 16.
    Johansson T, Bachrach-Lindström M, Aspenberg P, Jonsson D, Wahlström O (2006) The total costs of a displaced femoral neck fracture: comparison of internal fixation and total hip replacement. Int Orthop 30(1):1–6Google Scholar
  17. 17.
    Haentjens P, Lamraski G, Boonen S (2005) Costs and consequences of hip fracture occurrence in old age: an economic perspective. Disabil Rehabil 27(18–19):1129–1141Google Scholar
  18. 18.
    Pirson M, Leclercq P (2014) Un projet pilote d’évaluation des coûts par pathologie, le projet PACHA. Healthc Exec 78:12–14Google Scholar
  19. 19.
    Service Public Fédéral, Santé Publique, Sécurité de la Chaîne Alimentaire et Environnement, Organisation des établissements de soins, UZ Leuven (2017) Rapport project team APR-DRG I. La comparaison et la signification des résultats du grouper entre les versions APR-DRG 15.0 et 28.0 sur la base d’ICD-9-CM. Louvain; 2013. https://www.health.belgium.be/sites/default/files/uploads/fields/fpshealth_theme_file/icd-10-be_rapport_de_recherche_-_2012.pdf. Accessed 30 October 2017
  20. 20.
    Quan H, Sundararajan V, Halfon P, Fong A (2005) Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 43(11):1130–1139Google Scholar
  21. 21.
    Charlson ME, Pompei P, Ales KL, MacKenzie CR (1987) A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 40(5):373–383Google Scholar
  22. 22.
    Ruiz M, Bottle A, Long S, Aylin P (2015) Multi-morbidity in hospitalised older patients: who are the complex elderly? PLoS ONE 10(12):e0145372Google Scholar
  23. 23.
    Padrón-Monedero A, López-Cuadrado T, Galán I, Martínez-Sánchez EV, Martin P, Fernández-Cuenca R (2017) Effect of comorbidities on the association between age and hospital mortality after fall-related hip fracture in elderly patients. Osteoporos Int 28(5):1559–1568Google Scholar
  24. 24.
    De Foor J, Leclercq P, Van Den Bulcke J, Pirson M (2018) Impact des patients « bed blockers » sur les coûts hospitaliers et évaluation des obstacles à la sortie, étude prospective au sein de quatre hôpitaux belges. J Gest D’Écon Méd 35(4–5):179–196Google Scholar
  25. 25.
    Autier P, Haentjens P, Bentin J, Baillon JM, Grivegnée AR, Closon MC et al (2000) Costs induced by hip fractures: a prospective controlled study in Belgium. Osteoporos Int 11(5):373–380Google Scholar
  26. 26.
    Bozic KJ, Katz P, Cisternas M, Ono L et al (2005) Hospital resource utilization for primary and revision total hip arthroplasty. J Bone Jt Surg 87(3):570–576Google Scholar
  27. 27.
    De Block M (2015) Réforme du financement des hôpitaux. Association belge des hôpitaux. http://www.hospitals.be/images_produit/upload_fichier/334_2015-Congress-FR-Maggie-De-Blok.pdf. Accessed 22 August 2016
  28. 28.
    Scheerlinck T (2005) Meer revalidatiecentra kunnen kostprijs van heupfractuur en heupprothesechirurgie verminderen. Vrije Universiteit BrusselGoogle Scholar
  29. 29.
    de Béthune X, Ackaert K (2015) Étude MC prothèse totale de hanche - 24 ans d’analyses : l’étape suivante d’une collaboration transparente et équilibrée entre soignants et soignés et entre hôpitaux et mutualités. Mutualité Chrétienne. https://www.mc.be/actualite/communique-presse/2015/prothese-hanche#
  30. 30.
    Ackaert K, de Béthune X, Mertens R (2009) Prothèses totales de hanche en Belgique : analyse de suivi. Partie 2. Variation des types de prothèses et durée de survie. MC-inform 238:3–18Google Scholar
  31. 31.
    Pioli G, Bendini C, Pignedoli P, Giusti A, Marsh D (2018) Orthogeriatric co-management—managing frailty as well as fragility. Injury 49(8):1398–1402Google Scholar
  32. 32.
    Sabharwal S, Wilson H (2015) Orthogeriatrics in the management of frail older patients with a fragility fracture. Osteoporos Int 26(10):2387–2399Google Scholar
  33. 33.
    Petermans J (2014) Place du gériatre dans la filière de soins. Rev Méd Liège 69:5–6Google Scholar
  34. 34.
    Pepersack T (2013) Ortho-geriatrie: les preuves du concept. Rev Méd Brux 34(1):38–45Google Scholar
  35. 35.
    Tan L, Wong S, Kwek E (2017) Inpatient cost for hip fracture patients managed with an orthogeriatric care model in Singapore. Singapore Med J 58(3):139–144Google Scholar
  36. 36.
    Pueyo-Sánchez M-J, Larrosa M, Surís X, Sánchez-Ferrin P, Bullich-Marin I, Frigola-Capell E et al (2018) Association of orthogeriatric services with long-term mortality in patients with hip fracture. Eur Geriatr Med 9(2):175–181Google Scholar

Copyright information

© European Geriatric Medicine Society 2018

Authors and Affiliations

  • Julie De Foor
    • 1
    • 2
    Email author
  • Philippe Van Wilder
    • 2
  • Pol Leclercq
    • 2
  • Dimitri Martins
    • 2
  • Magali Pirson
    • 2
  1. 1.ICHEC Brussels Management SchoolBrusselsBelgium
  2. 2.Centre de recherche en Economie de la Santé (Health Economics Research Center), Gestion des Institutions de Soins et Sciences Infirmières (Management of Institutions of care and nursing research), Ecole de Santé Publique (School of Public Health), Université Libre de BruxellesBrusselsBelgium

Personalised recommendations