Archives of Osteoporosis

, 11:13

Time trends in hospital stay after hip fracture in Canada, 2004–2012: database study

  • Boris Sobolev
  • Pierre Guy
  • Katie Jane Sheehan
  • Lisa Kuramoto
  • Eric Bohm
  • Lauren Beaupre
  • Jason M. Sutherland
  • Michael Dunbar
  • Donald Griesdale
  • Suzanne N. Morin
  • Edward Harvey
  • for The Canadian Collaborative Study on Hip Fractures
Original Article

DOI: 10.1007/s11657-016-0264-5

Cite this article as:
Sobolev, B., Guy, P., Sheehan, K.J. et al. Arch Osteoporos (2016) 11: 13. doi:10.1007/s11657-016-0264-5

Abstract

Summary

Changes in bed management and access policy aimed to shorten Canadian hip fracture hospital stay. Secular trends in hip fracture total, preoperative, and postoperative stay are unknown. Hip fracture stay shortened from 2004 to 2012, mostly from shortening postoperative stay. This may reflect changes in bed management rather than access policy.

Purpose

To compare the probability of discharge by time after patient admission to hospital with first-time hip fracture over a period of nine calendar years.

Methods

We retrieved acute hospitalization records for 169,595 patients 65 years and older, who were admitted to an acute care hospital with hip fracture between 2004 and 2012 in Canada (outside of Quebec). The main outcome measure was cumulative incidence of discharge by inpatient day, accounting for competing events that end hospital stay.

Results

The probability of surgical discharge within 30 days of admission increased from 57.2 % in 2004 to 67.3 % in 2012. The probability of undergoing surgery on day of admission or day after fluctuated around 58.5 % over the study period. For postoperative stay, the discharge probability increased from 6.8 to 12.2 % at day 4 after surgery and from 57.2 to 66.6 % at day 21 after surgery, between 2004 and 2012. The differences across years persisted after adjustment for characteristics of patients, fracture, comorbidity, treatment, type and timing of surgery, and access to care.

Conclusions

Hospital stay following hip fracture shortened substantially between 2004 and 2012 in Canada, mostly due to shortening of postoperative stays. Shorter hospital stays may reflect changes in bed management protocols rather than in access policy.

Keywords

Hip fracture Length of stay Postoperative stay Time trends Cumulative incidence Competing risks 

Introduction

Two initiatives have shaped the length of hospital stay following hip fracture during the last decade in Canada: changes in bed management protocols and changes in policy on access to hip fracture surgery. Since the early 2000s, health authorities have promoted the early discharge of patients from hospital to other settings such as outpatient or home-based care [1, 2, 3, 4]. For example, a practice of repatriating hip fracture patients to long-term care within 5 days of their surgical care was introduced in some Canadian provinces [1, 5]. In 2005, provincial and territorial ministers of health announced the 48-hour benchmark for surgical treatment after admission with hip fracture as part of the national commitment for the timely provision of medical services [6]. Proponents of early surgery argue that the mandated treatment period improves survival, reduces complications, and shortens postoperative stay [7, 8, 9, 10, 11]. Although the Canadian Institute for Health Information (CIHI) reports access indicators for hip fracture surgery in Canada since 2007 [12], time trends in total and postoperative stay following hip fracture have remained unreported.

In the present study, we compared the probability of discharge within a certain time post admission with hip fracture between 2004 and 2012. We used the CIHI Discharge Abstract Database that routinely collects administrative, clinical, and demographic data from hospital discharges (including transfers, deaths, discharge destinations, and sign-outs) [13]. In estimating the probability of discharge after surgical treatment, we accounted for the competing events that end hospital stay, such as in-hospital death or discharge after nonoperative management. Differences between years were adjusted for patient characteristics, fracture type, comorbidities, treatment, and circumstances of their access to care [14, 15]. The University of British Columbia Behavioural Research Ethics Board approved this study (H11-02611).

Methods

Study group

We obtained all discharge abstracts with diagnostic codes of hip fracture (International Classification of Disease [ICD]-9 820; ICD-10 S72.0, S72.1, S72.2) for patients 65 years and older hospitalized between April 1, 2003 and March 31, 2013 in Canada, outside of Quebec. The study group included 150,928 patients with single abstract and 38,043 patients with multiple abstracts. We combined contiguous abstracts with the same patient ID into one episode of acute hospitalization when they were linked by hospital transfers before definitive care [16] (see Additional file 1). The earliest (index) admission and the latest discharge were designated as the beginning and the end of care episodes. We selected patients hospitalized with first hip fracture between January 1, 2004 (to avoid possible readmissions) and December 31, 2012 (to allow 90 days of follow-up). After excluding 222 patients with in-hospital hip fracture, 169,595 patients were available for the analysis (135,021 patients with one hospital stay and 34,574 patients with multihospital stays linked by transfers).

Outcomes

In this paper, hospital stay refers to inpatient days from index admission to discharge, preoperative stay refers to inpatient days from index admission to surgery, and postoperative stay refers to inpatient days between surgery and discharge. In the analysis of hospital stay, the outcome was discharge with surgical treatment; discharge without surgical treatment and in-hospital death were competing events (Fig. 1). Surgical procedures were identified with Canadian Classification of Intervention codes for hip fracture surgery in the discharge abstracts (1VA74^^, 1VA53^^, 1VC74^^, or 1SQ53^^). Live discharges were identified by the following codes: discharge to home without support services, discharge to home with support services, and transfer to long-term care or to nonacute care facility. Hospital stays were followed up to 30 days on the premise that longer stays reflect nonacute hospitalization [17]. In the analysis of preoperative stay, the primary event was surgery; discharge without surgical treatment and in-hospital death were competing events (Fig. 1). Preoperative stays were followed up to 21 days as longer delays may reflect patient’s unfitness for surgery [18]. In the analysis of postoperative stay, we did not include patients who die on the day of surgery. The primary event was hospital discharge, and in-hospital death was a competing event (Fig. 1). Postoperative stays were followed up to 21 days on the premise that this period represents acute hospitalization [11].
Fig. 1

Primary outcome, competing, and censoring events during total, preoperative, and postoperative stay. Solid lines indicate stays ended with primary outcome. Dashed lines indicate stays ended with competing events. Gray “x” indicate censoring events: transfer to another acute facility after definitive treatment, signed out against medical advice or did not return from pass, discharge occurring within 1 day of surgery, or the end of follow-up

Statistical analysis

Daily rates of discharge and surgery were estimated by dividing the number of corresponding events by the total number of inpatient days, overall and for each calendar year. The cumulative incidences of discharge and surgery as function of inpatient day were estimated by accounting for subhazards of respective competing events, for each calendar year [19]. We used proportional odds models to test whether the cumulative incidences of discharge and surgery were different in 2007, the year of the first report on hip fracture surgical times, as compared to other years [20]. We treated stays terminated by transfers to another acute care facility or by discharges occurring within 1 day of surgery as censored observations [21] (Fig. 1). The analysis was conducted with packages cmprsk [22], prodlim [23], and geepack [24] developed for R statistical environment, version 3.1.2 [25].

In the regression analysis, the differences between years were adjusted for age, sex, fracture type, comorbidity [26, 27], cancer potentially metastatic to bone, Paget’s disease of the bone, transfer history, province, hospital type (teaching, large, medium, or small community), weekday of admission, and time of admission. In the analysis of postoperative stay, we further adjusted for type of surgery [28] (fixation vs. implant) and timing of surgery (admission day or day after vs. three or more inpatient days). The literature suggests that age, type of fracture, comorbidity, and the need for preoperative cardiac testing may delay surgery [11, 29, 30]. Ho et al. related the length of postoperative stay to surgical delays [10]. Orosz et al. related surgical delays to system factors such as waiting for medical consultation or laboratory results, lack of an operating room or surgeon, or admission too late in the day [31]. In Canada, orthopedic care is commonly provided at large tertiary centers with transfer of patients from local hospitals. Tracey et al. noted that transferred patients were less likely to undergo surgery within two inpatient days than patients who were not transferred [32]. They also reported that patients admitted between midnight and noon or on a weekday were more likely to have surgery on the day of admission than patients admitted in the afternoon or evening or on a weekend. Ricci et al. reported that the weekday of hospital admission was an independent factor for time to surgery elsewhere [30].

Results

Patient characteristics

The majority of patients were women (72.5 %), 85 years of age or older (48.6 %) with transcervical fractures (51.6 %) (Table 1). Major comorbidity in the previous year was reported for 28.1 % patients, with cardiac dysrhythmias being the most prevalent (9.0 %). The distribution of patient characteristics differed across calendar years, with later years in the study progression seeing more men, 85–94 age group, subtrochanteric fractures, cardiac dysrhythmias, and hypertension but also less diabetes and acute or chronic ischemic heart disease (p < 0.001). Reported heart failure decreased to a low in 2008 but increased again by 2012 (p < 0.001). There were no differences across years in the prevalence of chronic obstructive pulmonary disease (p = 0.09), cancers potentially metastatic to bone (p = 0.19), or Paget’s disease of the bone (p = 0.12). Most hospitalizations occurred in the province of Ontario (48.4 %), in large community hospitals (41.3 %), without transfers before definitive care (91.6 %), on weekdays (72.4 %), and with admission after 6 pm (38.0 %). Internal fixation was the most common procedure (54.8 %). Later in the study period, there were fewer fixations, transfers, and admissions to small community hospitals (p < 0.001). Admission before 6 am was more prevalent in 2012 (p < 0.001). There were no differences across years for province (p = 0.16) or weekday of admission (p = 0.74).
Table 1

Characteristics of 169,595 patients with first hip fracture, by calendar year

Characteristic

All years (n = 169,595)

2004 (n = 19,090)

2005 (n = 19,114)

2006 (n = 18,548)

2007 (n = 18,626)

2008 (n = 18,584)

2009 (n = 18,703)

2010 (n = 18,730)

2011 (n = 18,762)

2012 (n = 19,438)

Age, years

 Mean (SD)

83.5 (7.7)

83.3 (7.5)

83.4 (7.5)

83.4 (7.6)

83.5 (7.6)

83.6 (7.7)

83.6 (7.7)

83.8 (7.7)

83.7 (7.9)

83.6 (8.0)

 65–74

23,553 (13.9)

2601 (13.6)

2626 (13.7)

2500 (13.5)

2509 (13.5)

2514 (13.5)

2589 (13.8)

2586 (13.8)

2667 (14.2)

2961 (15.2)

 75–84

63,694 (37.6)

7768 (40.7)

7573 (39.6)

7192 (38.8)

7151 (38.4)

6985 (37.6)

6882 (36.8)

6624 (35.4)

6724 (35.8)

6795 (35.0)

 85–94

71,341 (42.1)

7624 (39.9)

7780 (40.7)

7760 (41.8)

7799 (41.9)

7837 (42.2)

7929 (42.4)

8234 (44.0)

8017 (42.7)

8361 (43.0)

  ≥ 95

11,007 (6.5)

1097 (5.7)

1135 (5.9)

1096 (5.9)

1167 (6.3)

1248 (6.7)

1303 (7.0)

1286 (6.9)

1354 (7.2)

1321 (6.8)

Sexa

 Women

122,948 (72.5)

14,141 (74.1)

13,983 (73.2)

13,514 (72.9)

13,632 (73.2)

13,369 (71.9)

13,493 (72.1)

13,483 (72.0)

13,432 (71.6)

13,901 (71.5)

 Men

46,646 (27.5)

4949 (25.9)

5131 (26.8)

5034 (27.1)

4993 (26.8)

5215 (28.1)

5210 (27.9)

5247 (28.0)

5330 (28.4)

5537 (28.5)

Fracture typeb

 

 Transcervical

87,574 (51.6)

10,021 (52.5)

9879 (51.7)

9555 (51.5)

9538 (51.2)

9534 (51.3)

9687 (51.8)

9743 (52.0)

9584 (51.1)

10,033 (51.6)

 Pertrochanteric

72,546 (42.8)

8172 (42.8)

8311 (43.5)

8014 (43.2)

7998 (42.9)

8000 (43.0)

7968 (42.6)

7840 (41.9)

8038 (42.8)

8205 (42.2)

 Subtrochanteric

9475 (5.6)

897 (4.7)

924 (4.8)

979 (5.3)

1090 (5.9)

1050 (5.7)

1048 (5.6)

1147 (6.1)

1140 (6.1)

1200 (6.2)

Comorbidity

 Heart failurec

13,685 (8.1)

1578 (8.3)

1529 (8.0)

1447 (7.8)

1411 (7.6)

1366 (7.4)

1604 (8.6)

1552 (8.3)

1593 (8.5)

1605 (8.3)

 COPDc

9020 (5.3)

1026 (5.4)

1045 (5.5)

1059 (5.7)

955 (5.1)

1015 (5.5)

974 (5.2)

961 (5.1)

1008 (5.4)

977 (5.0)

 IHD (acute)c

10,222 (6.0)

1173 (6.1)

1218 (6.4)

1159 (6.2)

1186 (6.4)

1149 (6.2)

1054 (5.6)

1051 (5.6)

1153 (6.1)

1079 (5.6)

 Dysrhythmiasc

15,337 (9.0)

1512 (7.9)

1512 (7.9)

1456 (7.8)

1401 (7.5)

1572 (8.5)

1788 (9.6)

1867 (10.0)

2035 (10.8)

2194 (11.3)

 IHD (chronic)c

2842 (1.7)

396 (2.1)

367 (1.9)

329 (1.8)

297 (1.6)

283 (1.5)

308 (1.6)

272 (1.5)

316 (1.7)

274 (1.4)

 Hypertensionc

9550 (5.6)

1109 (5.8)

1131 (5.9)

1042 (5.6)

882 (4.7)

978 (5.3)

1001 (5.4)

1086 (5.8)

1142 (6.1)

1179 (6.1)

 Diabetesc

7522 (4.4)

365 (1.9)

411 (2.2)

878 (4.7)

1206 (6.5)

1310 (7.0)

983 (5.3)

779 (4.2)

789 (4.2)

801 (4.1)

 Cancerd,e

3503 (2.1)

362 (1.9)

373 (2.0)

362 (2.0)

403 (2.2)

378 (2.0)

382 (2.0)

401 (2.1)

427 (2.3)

415 (2.1)

 Paget’s diseasee

118 (0.1)

12 (0.1)

22 (0.1)

19 (0.1)

10 (0.1)

11 (0.1)

14 (0.1)

8 (0.0)

12 (0.1)

10 (0.1)

Transfer history

 

 Yes

14,234 (8.4)

1507 (7.9)

1670 (8.7)

1640 (8.8)

1634 (8.8)

1569 (8.4)

1615 (8.6)

1635 (8.7)

1596 (8.5)

1368 (7.0)

Province

 Newfoundland

3830 (2.3)

454 (2.4)

422 (2.2)

383 (2.1)

416 (2.2)

409 (2.2)

454 (2.4)

440 (2.3)

420 (2.2)

432 (2.2)

 Prince Edward Island

1216 (0.7)

130 (0.7)

144 (0.8)

120 (0.6)

154 (0.8)

129 (0.7)

138 (0.7)

132 (0.7)

137 (0.7)

132 (0.7)

 Nova Scotia

7198 (4.2)

809 (4.2)

827 (4.3)

745 (4.0)

781 (4.2)

761 (4.1)

847 (4.5)

799 (4.3)

803 (4.3)

826 (4.2)

 New Brunswick

5940 (3.5)

674 (3.5)

679 (3.6)

675 (3.6)

700 (3.8)

641 (3.4)

634 (3.4)

648 (3.5)

640 (3.4)

649 (3.3)

 Ontario

82,147 (48.4)

9355 (49.0)

9238 (48.3)

8949 (48.2)

9098 (48.8)

9034 (48.6)

8890 (47.5)

9013 (48.1)

9074 (48.4)

9496 (48.9)

 Manitoba

9709 (5.7)

1033 (5.4)

1110 (5.8)

1090 (5.9)

1041 (5.6)

1090 (5.9)

1110 (5.9)

1063 (5.7)

1100 (5.9)

1072 (5.5)

 Saskatchewan

8876 (5.2)

1081 (5.7)

1044 (5.5)

942 (5.1)

979 (5.3)

942 (5.1)

961 (5.1)

937 (5.0)

1000 (5.3)

990 (5.1)

 Alberta

18,560 (10.9)

2046 (10.7)

2084 (10.9)

2005 (10.8)

2022 (10.9)

1986 (10.7)

2090 (11.2)

2082 (11.1)

2071 (11.0)

2174 (11.2)

 British Columbia

31,849 (18.8)

3484 (18.3)

3537 (18.5)

3612 (19.5)

3406 (18.3)

3558 (19.1)

3546 (19.0)

3586 (19.1)

3478 (18.5)

3642 (18.7)

 Territories

270 (0.2)

24 (0.1)

29 (0.2)

27 (0.1)

29 (0.2)

34 (0.2)

33 (0.2)

30 (0.2)

39 (0.2)

25 (0.1)

Hospital type

 Teaching

59,150 (34.9)

6804 (35.6)

6744 (35.3)

6500 (35.0)

6474 (34.8)

6463 (34.8)

6533 (34.9)

6422 (34.3)

6541 (34.9)

6669 (34.3)

 Community-large

69,975 (41.3)

7383 (38.7)

7503 (39.3)

7253 (39.1)

7488 (40.2)

7710 (41.5)

7859 (42.0)

8028 (42.9)

8006 (42.7)

8745 (45.0)

 Community-medium

27,375 (16.1)

2972 (15.6)

3026 (15.8)

3125 (16.8)

3102 (16.7)

2994 (16.1)

2985 (16.0)

2994 (16.0)

3023 (16.1)

3154 (16.2)

 Community-small

8526 (5.0)

1021 (5.3)

1020 (5.3)

996 (5.4)

1027 (5.5)

984 (5.3)

945 (5.1)

893 (4.8)

835 (4.5)

805 (4.1)

 Unknown

4569 (2.7)

910 (4.8)

821 (4.3)

674 (3.6)

535 (2.9)

433 (2.3)

381 (2.0)

393 (2.1)

357 (1.9)

65 (0.3)

Admission day

 Weekday

122,799 (72.4)

13,803 (72.3)

13,860 (72.5)

13,387 (72.2)

13,578 (72.9)

13,455 (72.4)

13,482 (72.1)

13,582 (72.5)

13,628 (72.6)

14,024 (72.1)

 Weekend

46,796 (27.6)

5287 (27.7)

5254 (27.5)

5161 (27.8)

5048 (27.1)

5129 (27.6)

5221 (27.9)

5148 (27.5)

5134 (27.4)

5414 (27.9)

Admission time

 Unknown

94 (0.1)

0 (0.0)

0 (0.0)

0 (0.0)

0 (0.0)

0 (0.0)

0 (0.0)

0 (0.0)

55 (0.3)

39 (0.2)

 00:00–5:59

23,909 (14.1)

2599 (13.6)

2632 (13.8)

2478 (13.4)

2608 (14.0)

2766 (14.9)

2603 (13.9)

2689 (14.4)

2644 (14.1)

2890 (14.9)

 6:00–11:59

23,792 (14.0)

2574 (13.5)

2508 (13.1)

2532 (13.7)

2538 (13.6)

2673 (14.4)

2770 (14.8)

2693 (14.4)

2709 (14.4)

2795 (14.4)

 12:00–17:59

57,343 (33.8)

6611 (34.6)

6476 (33.9)

6389 (34.4)

6380 (34.3)

6191 (33.3)

6271 (33.5)

6357 (33.9)

6273 (33.4)

6395 (32.9)

 18:00–23:59

64,457 (38.0)

7306 (38.3)

7498 (39.2)

7149 (38.5)

7100 (38.1)

6954 (37.4)

7059 (37.7)

6991 (37.3)

7081 (37.7)

7319 (37.7)

Type of surgery

 Not applicable

14,330 (8.4)

1667 (8.7)

1585 (8.3)

1570 (8.5)

1511 (8.1)

1612 (8.7)

1578 (8.4)

1639 (8.8)

1607 (8.6)

1561 (8.0)

 Internal fixation

92,887 (54.8)

10,549 (55.3)

10,818 (56.6)

10,239 (55.2)

10,346 (55.5)

10,128 (54.5)

10,120 (54.1)

10,163 (54.3)

10,154 (54.1)

10,370 (53.3)

 Implant/other

62,378 (36.8)

6874 (36.0)

6711 (35.1)

6739 (36.3)

6769 (36.3)

6844 (36.8)

7005 (37.5)

6928 (37.0)

7001 (37.3)

7507 (38.6)

SD standard deviation, COPD chronic obstructive pulmonary disease, IHD ischemic heart disease

aOne patient with unknown sex

bAt admission. For 2074 patients with different fracture types at admission and surgery, the fracture type at surgery is presented

cBy diagnostic codes from all hospitalizations in 1 year prior to index admission

dIncludes breast–female, prostate, renal, lung, multiple myeloma, and metastatic cancer

eBy using diagnostic codes from all hospitalizations during the hip fracture care episode

Inpatient days during acute hospitalization

In total, 169,595 hip fracture patients stayed in hospitals for 2,578,520 days, with a median stay of 12 days (interquartile range [IQR] 7–23 days). The number of days varied from 274,588 days in 2006 to 296,082 days in 2012. Overall, 155,265 (91.6 %) patients underwent surgical treatment and 14,330 (8.5 %) did not have surgery and 10,736 (6.9 %) surgical and 4143 (28.9 %) nonsurgical patients died in hospital. Among those who underwent surgery, the median time to surgery was 2 days (IQR 2–3 days).

The hospital stay of 40,787 (24.0 %) patients included alternate level of care (ALC) days, i.e., time patients remained in hospital awaiting placement into extended care units, nursing homes, or home care programs [33]. Among patients with ALC days, 36,149 (88.6 %) were discharged following surgical treatment, 2485 (6.1 %) were discharged without surgical treatment, and 2153 (5.3 %) died in hospital. Among the 128,808 patients without ALC days, 108,380 (84.1 %) were discharged following surgical treatment, 7702 (6.0 %) were discharged without surgical treatment, and 12,726 (9.9 %) died in hospital.

Total hospital stay

By inpatient day 30, 93,623 (55.2 %) hospital stays ended with discharge with surgery, 11,919 (7.0 %) with in-hospital death, and 6669 (3.9 %) with discharge without surgery. Overall, 57,384 (33.8 %) stays were terminated by censoring events or by the end of follow-up. The average rate of discharge was 3.6 (95 % confidence interval [CI] 3.6 to 3.7) per 100 patient-days, increasing from 2.9 (95 % CI 2.9 to 3.0) in 2004 to 4.1 (95 % CI 4.0 to 4.1) in 2012 (Table 2). The cumulative incidence of discharge at inpatient day 8 increased from 15.0 % in 2004 to 23.5 % in 2012 and at day 30 from 57.2 % in 2004 to 67.3 % in 2012 (Fig. 2). Compared to 2007, the odds of surgical discharge were 30 % lower in 2004, odds ratio [OR] = 0.70 (95 % CI 0.67 to 0.73), and 24 % lower in 2005, OR = 0.76 (95 % CI 0.73 to 0.79), but then 7 % higher in 2011, OR = 1.07 (95 % CI 1.03 to 1.12), and 16 % higher in 2012, OR = 1.16 (95 % CI 1.12 to 1.21) (Table 2).
Table 2

Cumulative incidence function of surgical discharge for all patients admitted with first hip fracture, by calendar year

Year

No. of patients

No. of dischargesa

Discharge rate (95 % CI)b

30-day CIF, % (95 % CI)

Crude OR of CIF (95 % CI)

Adjusted OR of CIF (95 % CI)c,d

2004

19,090

8305

2.9 (2.9–3.0)

57.2 (56.4–58.1)

0.74 (0.72–0.77)

0.70 (0.67–0.73)

2005

19,114

8813

3.1 (3.0–3.2)

59.1 (58.3–59.9)

0.79 (0.76–0.82)

0.76 (0.73–0.79)

2006

18,548

9968

3.6 (3.6–3.7)

64.0 (63.3–64.8)

0.96 (0.92–0.99)

0.95 (0.92–0.99)

2007

18,626

10,469

3.7 (3.7–3.8)

64.9 (64.2–65.6)

1.00

1.00

2008

18,584

10,746

3.7 (3.7–3.8)

64.5 (63.7–65.2)

0.99 (0.95–1.03)

0.99 (0.95–1.03)

2009

18,703

10,948

3.7 (3.7–3.8)

64.8 (64.1–65.5)

0.99 (0.95–1.03)

1.00 (0.96–1.04)

2010

18,730

11,080

3.8 (3.7–3.9)

65.0 (64.2–65.7)

1.02 (0.98–1.06)

1.04 (1.00–1.08)

2011

18,762

11,266

3.9 (3.8–3.9)

65.7 (65.0–66.4)

1.05 (1.01–1.09)

1.07 (1.03–1.12)

2012

19,438

12,028

4.1 (4.0–4.1)

67.3 (66.6–68.0)

1.14 (1.10–1.19)

1.16 (1.12–1.21)

CIF cumulative incidence function, CI confidence interval, OR odds ratio

aAt 30 days from admission

bPer 100 patient-days

cAdjusted for age, sex, fracture type, comorbidity, transfer history, province, hospital type, weekday of admission, and time of admission. CIF pseudovalues are calculated at inpatient day 3, 4, 6, 8, 12, 16, 20, 24, and 30

dExcludes 101 patients with unknown date of discharge, sex, or time at admission

Fig. 2

Cumulative incidence of surgical discharge by inpatient day across years

Preoperative stay

By inpatient day 21, 154,607 (91.2 %) hospital stays ended with surgery, 5790 (3.4 %) with discharge without surgery, and 3567 (2.1 %) with in-hospital death. In total, 5631 (3.3 %) stays were terminated by censoring events or by the end of follow-up. Overall, 20.1 % patients underwent surgery on the day of admission (inpatient day 1). This proportion decreased between 2004 and 2012 from 22.0 to 19.3 % (Fig. 3). The cumulative incidence of surgery at inpatient day 2 decreased from 60.8 % in 2004 to 56.7 % in 2008 and then increased to 60.7 % in 2012 (p < 0.001). Its value at inpatient day 3 decreased from 78.0 % in 2004 to 76.2 % in 2008 but increased thereafter (Fig. 3). Compared to 2007, the odds of surgery were 16 % higher in 2004, OR = 1.16 (95 % CI 1.12 to 1.21), 13 % higher in 2005, OR = 1.13 (95 % CI 1.09 to 1.18), and then first marginally higher in 2010, OR = 1.05 (95 % CI 1.01 to 1.10), and 12 % higher in 2012, OR = 1.12 (95 % CI 1.07 to 1.16) (Table 3). Compared to the province of Ontario, the odds of surgery were 9 to 43 % higher in other provinces (Additional file 2).
Fig. 3

Cumulative incidence of surgery by inpatient day across years

Table 3

Cumulative incidence function of surgery for all patients admitted with first hip fracture, by calendar year

Year

No. of patients

No. of surgeriesa

Surgery rate (95 % CI)b

21-day CIF, % (95 % CI)

Crude OR of CIF (95 % CI)

Adjusted OR of CIF (95 % CI)c,d

2004

19,090

17,356

27.3 (26.9–27.7)

91.5 (91.1–91.9)

1.16 (1.12–1.21)

1.16 (1.12–1.21)

2005

19,114

17,454

27.2 (26.8–27.6)

91.8 (91.4–92.1)

1.11 (1.07–1.15)

1.13 (1.09–1.18)

2006

18,548

16,889

26.8 (26.4–27.2)

91.4 (91.1–91.8)

1.02 (0.98–1.06)

1.03 (0.99–1.07)

2007

18,626

17,030

27.1 (26.7–27.5)

91.9 (91.5–92.2)

1.00

1.00

2008

18,584

16,890

26.8 (26.4–27.2)

91.3 (90.9–91.7)

0.99 (0.95–1.03)

0.99 (0.95–1.03)

2009

18,703

17,057

27.3 (26.9–27.7)

91.5 (91.1–91.9)

1.01 (0.97–1.05)

1.03 (0.99–1.07)

2010

18,730

17,031

27.1 (26.6–27.5)

91.2 (90.8–91.6)

1.04 (1.00–1.08)

1.05 (1.01–1.10)

2011

18,762

17,088

27.3 (26.9–27.8)

91.3 (90.9–91.7)

1.03 (0.99–1.07)

1.04 (1.00–1.08)

2012

19,438

17,812

28.5 (28.1–28.9)

91.9 (91.5–92.2)

1.14 (1.09–1.18)

1.12 (1.07–1.16)

CIF cumulative incidence function, CI confidence interval, OR odds ratio

aAt 21 days from admission

bPer 100 patient-days

cAdjusted for age, sex, fracture type, comorbidity, transfer history, province, hospital type, weekday of admission, and time of admission. CIF pseudovalues are calculated at inpatient day 1, 2, 3, 4, 5, 6, 7, 10, 14, and 21

dExcludes 100 patients with unknown procedure date, sex, or time at admission

Postoperative stay

During the 21-day period after surgery, 86,332 (55.7 %) stays ended with discharge and 7316 (4.7 %) with in-hospital death. Overall, 61,299 (39.6 %) stays were terminated by censoring events or by the end of follow-up. After surgery, the discharge rate was 4.8 (95 % CI 4.8 to 4.8) per 100 patient-days, increasing from 3.9 (95 % CI 3.8 to 4.0) in 2004 to 5.3 (95 % CI 5.2 to 5.4) in 2012 (Table 4). The cumulative incidence of discharge increased between 2004 and 2012 from 6.8 to 12.2 % at day 4 after surgery and from 57.2 to 66.6 % at day 21 after surgery (Fig. 4). Compared to 2007, the odds of discharge were 32 % lower in 2004, OR = 0.68 (95 % CI 0.65 to 0.71), 26 % lower in 2005, OR = 0.74 (95 % CI 0.71 to 0.77), but then 7 % higher in 2011, OR = 1.07 (95 % CI 1.02 to 1.11), and 16 % higher in 2012, OR = 1.16 (95 % CI 1.11 to 1.21) (Table 4). Patients undergoing surgery after two inpatient days had 23 % lower odds of discharge than patients who had surgery on day of admission or day after, OR = 0.77 (95 % CI 0.74 to 0.79) (Additional file 2). Compared to the province of Ontario, the odds of discharge were 44 to 65 % lower in other provinces (Additional file 2).
Table 4

Cumulative incidence function of discharge for surgically treated patients with first hip fracture, by calendar year

Year

No. of patients

No. of dischargesa

Discharge rate (95 % CI)b

21-day CIF, % (95 % CI)

Crude OR of CIF (95 % CI)

Adjusted OR of CIF (95 % CI)c,d

2004

17,388

7639

3.9 (3.8–4.0)

57.2 (55.5–58.9)

0.72 (0.69–0.75)

0.68 (0.65–0.71)

2005

17,491

8145

4.1 (4.0–4.2)

56.7 (55.8–57.5)

0.77 (0.74–0.80)

0.74 (0.71–0.77)

2006

16,939

9203

4.8 (4.7–4.9)

59.8 (59.0–60.7)

0.95 (0.92–0.99)

0.95 (0.91–0.99)

2007

17,077

9656

4.9 (4.8–5.0)

65.1 (64.4–65.9)

1.00

1.00

2008

16,943

9905

4.9 (4.8–5.0)

65.4 (64.6–66.1)

0.99 (0.95–1.03)

1.00 (0.96–1.04)

2009

17,085

10,059

4.9 (4.8–5.0)

64.2 (63.4–64.9)

0.98 (0.94–1.02)

0.99 (0.95–1.04)

2010

17,052

10,227

5.1 (5.0–5.1)

65.3 (64.5–66.1)

1.03 (0.99–1.07)

1.06 (1.01–1.10)

2011

17,127

10,391

5.1 (5.0–5.2)

65.7 (64.9–66.4)

1.05 (1.01–1.09)

1.07 (1.02–1.11)

2012

17,845

11,107

5.3 (5.2–5.4)

66.6 (65.8–67.3)

1.13 (1.08–1.18)

1.16 (1.11–1.21)

CIF cumulative incidence function, CI confidence interval, OR odds ratio

aAt 21 days after surgery

bPer 100 patient-days

cAdjusted for age, sex, fracture type, comorbidity, transfer history, province, hospital type, weekday of admission, time of admission, type of surgery, and timing of surgery. CIF pseudovalues are calculated at inpatient days 2, 3, 4, 5, 6, 7, 10, 14, and 21

dExcludes 101 patients with unknown procedure date, discharge date, sex or time at admission

Fig. 4

Cumulative incidence of discharge by day after surgery across years

Discussion

We compared hospital stays of patients hospitalized with first-time hip fracture in Canada for calendar years between 2004 and 2012. This period saw changes in bed management and introduction of the 48-hour benchmark for hip fracture surgery. We found that the daily rate of surgical discharge increased between 2004 and 2012 from 2.9 to 4.1 per 100 patient-days, and the probability of surgical discharge within 30 days of admission increased from 57.2 % in 2004 to 67.3 % in 2012, indicating that the hospital stay shortened substantially over the study period. This change was mostly driven by shorter postoperative stay, with the probability of discharge increasing from 6.8 to 12.2 % at day 4 after surgery and from 57.2 to 66.6 % at day 21 after surgery. During the study period, the probability of undergoing early surgery (on day of admission or day after) remained essentially unchanged. The observed differences across the calendar years persisted after adjustment for characteristics of patients, fracture type, comorbidity, treatment, type and timing of surgery, and access to care, with the odds ratios for the adjustment covariates differing from 1 (Additional file 2).

In recent years, hospital stays for surgical treatment shortened in Canada [34, 35, 36] and elsewhere [17, 37]. The degree of shortening varied across provinces in Canada (Additional file 2). This may be explained by the provincial responsibility for health care delivery in Canada, whereby each of ten provinces and territories control their own health care budget, allocations, planning, and infrastructure. The observed shortening of hospital stay in hip fracture patients in this study is consistent with other reports [27, 38, 39], except for one Australian report [40]. Similar to methods of previous reports, we used discharge abstracts to identify the dates of index admission, operation, and discharge from hospital and the type of discharge. To go beyond their methodology, we estimated the discharge probability using the cumulative incidence function that accounts for the risk of events competing for ending a hospital stay [19]. It is possible that focusing solely on the number of inpatient days overlooks that in-hospital death before surgery, discharge without surgical treatment, or return to long-term care may shorten the duration of stay in acute hospital [41].

Globally, several initiatives were introduced to reduce length of hospital stay: multidisciplinary team care, improved discharge planning, early supported discharge programs, and clinical care pathways [42]. In the UK hip fracture patients, reduction in preoperative stay was linked to a national audit program [43], and reduction in total hospital stay was linked to a higher rate of transfers to subacute care [38]. Similarly in Japan, shorter acute hospital stays were linked to increased number of available long-term care beds [33]. A shorter hospital stay following hip fracture was linked to admission to a consolidated geriatric care unit [44] and to a co-managed care pathway [45]. At the same time, some researchers submitted that patient-related factors may outweigh changes in care models for those who require more time for a safe discharge remaining longer in hospital [40]. Based on these results, we propose that shorter hospital stays have reflected changes in bed management rather than access to surgery because those changes target different segments of the hospital stay. The lack of differences in the preoperative stay is most likely related to challenges in prioritizing hip fracture surgery over other urgent surgical procedures within one hospital.

Patients admitted with hip fracture are identified as surgical or nonsurgical candidates. Most patients admitted with first hip fracture underwent surgery (91.6 %). Those that were discharged without surgery (3.4 %) may represent patients who were nonambulatory and institutionalized [46], with stable fractures [46], with a high risk of surgical complications [47], or who transferred to palliative care [48]. These surgical and nonsurgical pathways impact length of stay. We accounted for this in our analysis with live nonoperative discharge as a competing event. Nonsurgical patients returned to long-term care are likely to have a shorter length of stay than surgical patients returned to the community. Future study on the impact of discharge disposition on length of stay would shed further light on the declining secular trends noted in the current study.

We conducted an analysis of routinely collected discharge abstracts with a limited number of variables for adjustment. In particular, no information was available on socio-economic status, family environment and support, previous dementia, or degree of independence for activities of daily living which may impact length of hospital stay. Furthermore, no information was available on the rate of postoperative complications among these patients, which may have further impacted length of hospital stay. Data on hip fractures sustained before 2003 were not available. Therefore, we excluded all patients admitted before 2004 to avoid misclassifying possible readmissions or subsequent hip fractures as hospitalization for first-time hip fracture. Due to limited clinical information, we could not distinguish nonsurgical candidates from surgical candidates who die before surgery. However, the use of the cumulative incidence function of discharge allowed us to account for admissions ended without surgery, either due to nonoperative management or death before surgery. A number of factors could have affected length of stay in individual hospitals during the study period, such as the timing of new policies around access to surgery, management initiatives to reduce ALC, and changes to the methods of funding hospitals. However, the extent to which these regional differences play a role, alone or jointly, in the observed time trends is beyond the scope of this study. We only adjusted for the type of hospital where the patient was treated as a proxy for standards of anesthesia, surgery, and intensive care; adequacy of facilities and staffing levels; attitude to training; and interpersonal relationships among staff.

Acknowledgments

We gratefully acknowledge the guidance from the CIHI experts in understanding the discharge abstracts.

Compliance with ethical standards

Funding source

This research was funded by the Canadian Institute for Health Research. This funder had no role in the design of this study, execution, analyses, data interpretation, or decision to submit results for publication.

Conflicts of interest

The following competing interests are declared: (1) PG has received grants from the Canadian Institutes of Health Research related to this work. PG also receives funding from the Natural Sciences and Engineering Research Council of Canada, the Canadian Foundation for Innovation, and the British Columbia Specialists Services Committee for work around hip fracture care not related to this manuscript. He has also received fees from the BC Specialists Services Committee (for a provincial quality improvement project on redesign of hip fracture care) and from Stryker Orthopedics (as a product development consultant). He is a board member and shareholder in Traumis Surgical Systems Inc. and a board member for the Canadian Orthopedic Foundation. He also serves on the speakers’ bureaus of AO Trauma North America and Stryker Canada. (2) SNM reports grants from Amgen Canada, grants from Merck, personal fees from Amgen Canada, and personal fees from Eli Lilly outside the submitted work. (3) KS is a postdoctoral fellow whose salary is paid by Canadian Institutes of Health Research funding related to this work. (4) BS, LK, EB, LB, JMS, MD, DG, EH declare that they have no conflicts of interest.

Supplementary material

11657_2016_264_MOESM1_ESM.doc (30 kb)
Additional File 1(DOC 30 kb)
11657_2016_264_MOESM2_ESM.doc (190 kb)
Additional File 2(DOC 189 kb)

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2016

Authors and Affiliations

  • Boris Sobolev
    • 1
  • Pierre Guy
    • 2
  • Katie Jane Sheehan
    • 1
  • Lisa Kuramoto
    • 3
  • Eric Bohm
    • 4
  • Lauren Beaupre
    • 5
  • Jason M. Sutherland
    • 1
  • Michael Dunbar
    • 6
  • Donald Griesdale
    • 7
  • Suzanne N. Morin
    • 8
  • Edward Harvey
    • 9
  • for The Canadian Collaborative Study on Hip Fractures
  1. 1.School of Population and Public HealthUniversity of British ColumbiaVancouverCanada
  2. 2.Department of OrthopedicsUniversity of British ColumbiaVancouverCanada
  3. 3.Vancouver Coastal Health Research InstituteUniversity of British ColumbiaVancouverCanada
  4. 4.Division of Orthopaedic Surgery and Center for Healthcare InnovationUniversity of ManitobaWinnipegCanada
  5. 5.Departments of Physical Therapy and Division of Orthopaedic SurgeryUniversity of AlbertaEdmontonCanada
  6. 6.Division of Orthopaedic SurgeryDalhousie UniversityHalifaxCanada
  7. 7.Department of Anesthesiology, Pharmacology & TherapeuticsUniversity of British ColumbiaVancouverCanada
  8. 8.Department of MedicineMcGill UniversityMontréalCanada
  9. 9.Division of Orthopaedic SurgeryMcGill UniversityMontréalCanada

Personalised recommendations