Abstract
Aims
To compare 30-day survival and recovery of prefracture ambulation between patients mobilised early (on the day of or day after surgery) and patients mobilised late (2 days of more after surgery) in England and Wales. To determine whether the presence of dementia influences the association between mobilisation timing and 30-day survival and recovery.
Methods
Analysis of the National Hip Fracture Database linked to hospitalisation records for 126,897 patients ≥60 years surgically treated for hip fracture in England/Wales between 2014 and 2016. Using logistic regression, we adjusted for covariates with a propensity score to estimate the association between mobilisation timing and survival and ambulation recovery.
Results
99,667 (79%) patients mobilised early. Among those mobilised early compared to those mobilised late, the weighted odds ratio of survival was 1.92 (95% CI 1.80 – 2.05), of recovering outdoor ambulation was 1.25 (95% CI 1.03 – 1.51), and of recovering indoor ambulation was 1.53 (95% CI 1.32 – 1.78) by 30 days. The weighted probability (%) of survival at 30-days post-admission were 95.9 (95% CI 95.7 – 96.0) and 92.4 (95% CI 92.0 – 92.8) respectively among those mobilised early and those mobilised late. The weighted probability of ambulation recovery given outdoor and indoor ambulation prefracture were 9.7 (95% CI 9.2-10.2) and 81.2 (95% CI 80.0-82.4) respectively among those mobilised early, and were 7.9 (95% CI 6.6 – 9.2) and 73.8 (95% CI 71.3 – 76.2) respectively among those mobilised late. Patients with dementia were less likely to mobilise early despite observed associations with survival and ambulation recovery for those with and without dementia.
Conclusion
Early mobilisation was associated with survival and recovery for patients (with and without dementia) after hip fracture. Early mobilisation should be incorporated as a measured indicator of quality. Reasons for failure to mobilise early should also be captured to inform quality improvement initiatives.
Introduction
In the UK, 30-day survival after hip fracture increased from 88.5% in 2003 to 93.9% in 2018 as demonstrated by the National Hip Fracture Database (NHFD) with 95% case ascertainment.1
Improvements in survival have been attributed to improved clinical care associated with identification and audit of performance indicators, with feedback to participating sites.2 Recent indicators include early mobilisation in the expectation this impacts rates of survival and recovery.1
Evidence underlying recommendations for early mobilisation are based on one low-moderate quality trial of 60 patients.3 In the UK, further trials may not be ethical/feasible as early mobilisation was incorporated into usual care based on national guidance.3 However, analysis of real-world observational data presents an opportunity to study the association between early mobilisation and outcomes.4 Indeed, a cohort of 532 patients in New York points to a possible 6-month survival and 2-month recovery benefit of early mobilisation,5 while an analysis of unlinked NHFD data from 2013-2015 reported a 30-day recovery benefit.6 These analyses could be built upon with estimation of survival among a larger patient cohort, adjustment for confounders with the use of linked data, and by considering ambulation in light of pre-fracture ambulatory ability.
These benefits reflect mean (or median) values encompassing a wide range of patients. Whether all patients benefit remains unclear. Previous analyses reported patients with dementia are less likely to mobilise early7 and less likely to survive to 30-days.8 However, a subgroup analysis of a trial of rehabilitation in patients with hip fracture found those with mild to moderate dementia benefited more from the intervention in terms of independence at 3-months than those without or with severe dementia.9 This intervention included early mobilisation but the extent to which this aspect of care contributed to beneficial effects observed is unclear.9 Directly comparing the association between early mobilisation and outcomes across subgroups defined by dementia would resolve this uncertainty.
The objectives of this study were to compare 30-day survival and recovery of prefracture ambulation between patients mobilised early and those mobilised late; and to determine whether the presence of dementia influences the association between mobilisation timing and 30-day survival and recovery.
Methods
This study is reported according to the REporting of studies Conducted using Observational Routinely-collected Data statement.4 This study received NHS Health Research Authority and Health and Care Research Wales approval (IRAS Project ID: 230215). The study did not require ethical approval as an analysis of pseudonymized data.
Cohort
The NHFD assembles data on the characteristics of patients and their care following acute admission with hip fracture.1 Between January 1st 2014 and December 31st 2016, data were submitted for 170,970 patients ≥60 years surgically treated for nonpathological first hip fracture with a postoperative hospital stay of ≥1 day in England or Wales. Case ascertainment was estimated at 95% (for 2015).10 Data were linked to the English Hospital Episode Statistics database and Patient Episode Database for Wales for data on comorbidities, ethnicity, deprivation, and survival. Details of data cleaning, linkage, selection, and validation are available elsewhere.7 We selected patients with complete data for exposure and either of our primary outcomes (n = 126,897). Differences between patients with and without complete data are presented in Supplementary File 1, Table S4-S5.
Exposure
The exposure was an indicator for timing of mobilisation: ‘early’ (on the day of or day after surgery) or ‘late’ (2 days of more after surgery). The NHFD defines mobilisation by the ability to sit or stand out of bed with or without help.1
Outcomes
The primary outcome was survival at 30-days post-admission. The secondary outcome was ambulation recovery defined as no change (or improvement) in ambulation level from prefracture to 30-days post-admission. We defined levels of ambulation based on patient (or surrogate) report of ambulatory outdoors [NHFD: ambulatory in and outdoors without aids, ambulatory outdoors with one aid, ambulatory outdoors with two aids or frame] or ambulatory indoors [NHFD: some indoor ambulation but never goes outside without help]. We chose this definition to reflect ambulatory ability which increases opportunity for wider social participation. We excluded those with no ambulation prefracture from recovery analysis to prevent overestimation of recovery.
Subgroup
We used ICD-10 codes to identify patients with dementia [ICD-10: E100-E108, E110-E118, E130-E138, E140-E148] during their hip fracture admission or an admission in the year prior.
For recovery analysis, we stratified patients according to prefracture ambulation: outdoors or indoors only.
Statistical analysis
We used Stata version 16 for analysis.11 We described patient and care characteristics by median and interquartile ranges for continuous variables and proportions for categorical variables, overall and by mobilisation timing. We used the χ2 test and the Wilcoxon rank test to compare distributions by mobilisation timing. We estimated the proportion of patients mobilised early and survived and recovered by 30-days postadmission.
We defined a propensity score (PS) for mobilisation timing with respect to confounders using a logistic regression and defined weights to estimate the average treatment effect, equal to 1/PS if a patient was mobilised early and 1/(1-PS) otherwise with Stata psmatch212 and scenttest13 (Supplementary File 2).14 We used logistic regression with propensity score weighting to regress survival and ambulation recovery at 30-days with respect to mobilisation timing, overall and by dementia. We summarised results with odds ratios and probabilities.
Sensitivity analysis
We assessed the influence of missing data on mobilisation timing -survival association through multiple imputation by chained equations15 using mi impute chained command in Stata version 16.11 We replaced missing values with a random sample of imputed values to generate 50 distinct datasets via imputation models for each variable based on a rule to reduce sampling variability while limiting loss of power for assessing the association to no more than 1% (Supplementary File 3, Table S1).15,16 We used Rubin’s rules to obtain the combined point estimate of the odds ratios from 50 datasets.17 Propensity scores were estimated post imputation for each of the 50 datasets using Within and Across approaches.18,19 We did not explore the potential influence of missing data in our analysis of 30-day recovery as the data was not missing at random.15
Results
Patient characteristics
Of the 126,897 patients, 72% were female, 78% white, 55% with an ASA grade III, 79% admitted from home, and 38% ambulatory in and outdoors without aids prefracture (Table 1).
Table 1. Characteristics of patients surgically treated for non-pathological first hip fracture overall and by timing of mobilisation in the complete case analysis dataset of day 30 survival outcome (n = 126,897).
| All (n = 126,897) | early mobilisation (n = 99,667) | delayed mobilisation (n = 27230) | ||
|---|---|---|---|---|
| Age (years) Median (IQR) (n=l26,665) | 84 [77-89] | 84 [77-89] | 85 [79-90] | |
| Sex | Female | 91962 (72.5) | 72650 (79.0) | 19312 (21.0) |
| Male | 34933 (27.5) | 27016 (77.3) | 7917 (22.7) | |
| Missing | 2 (0.0) | 1 (50.0) | 1 (50.0) | |
| Ethnicity | White | 99585 (78.5) | 78948 (79.3) | 20637 (20.7) |
| Caribbean or African or any mixed black background | 244 (0.2) | 159 (65.2) | 85 (34.8) | |
| Asian or Asian British or any mixed Asian background | 1262 (1.0) | 970 (76.9) | 292 (23.1) | |
| Missing | 25806 (20.3) | 19590 (75.9) | 6216 (24.1) | |
| Ambulatory ability prior to hip fracture | Freely ambulatory without aids | 47990 (37.8) | 40516 (84.4) | 7474 (15.6) |
| Ambulatory outdoors with one aid | 28218 (22.2) | 22632 (80.2) | 5586 (19.8) | |
| Ambulatory outdoors with two aids or frame | 17929 (14.1) | 13813 (77.0) | 4116 (23.0) | |
| Some indoor ambulation but never goes outside without help | 29567 (23.3) | 20660 (69.9) | 8907 (30.1) | |
| No functional ambulation | 1717 (1.4) | 1016 (59.2) | 701 (40.8) | |
| Missing | 1476 (1.2) | 1030 (69.8) | 446 (30.2) | |
| Deprivation | least deprived 10% | 10545 (8.3) | 8194 (77.7) | 2351 (22.3) |
| less deprived 10-20% | 10335 (8.1) | 7939 (76.8) | 2396 (23.2) | |
| less deprived 20-30% | 11254 (8.9) | 8650 (76.9) | 2604 (23.1) | |
| less deprived 30-40% | 12023 (9.5) | 9342 (77.7) | 2681 (22.3) | |
| less deprived 40-50% | 12618 (9.9) | 9874 (78.3) | 2744 (21.7) | |
| more deprived 40-50% | 13315 (10.5) | 10430 (78.3) | 2885 (21.7) | |
| more deprived 30-40% | 13149 (10.4) | 10337 (78.6) | 2812 (21.4) | |
| more deprived 20-30% | 12701 (10.0) | 10068 (79.3) | 2633 (20.7) | |
| more deprived 10-20% | 12622 (10.0) | 10095 (80.0) | 2527 (20.0) | |
| most deprived 10% | 11910 (9.4) | 9587 (80.5) | 2323 (19.5) | |
| Missing | 6425 (5.1) | 5151 (80.2) | 1274 (19.8) | |
| Hip fracture type | Intracapsular | 74886 (59.0) | 59322 (79.2) | 15564 (20.8) |
| Intertrochanteric | 44463 (35.0) | 34800 (78.3) | 9663 (21.7) | |
| Subtrochanteric | 7488 (5.9) | 5499 (73.4) | 1989 (26.6) | |
| Missing | 60 (0.1) | 46 (76.7) | 14 (23.3) | |
| Surgery timing | Within target time | 90713 (71.5) | 72127 (79.5) | 18586 (20.5) |
| Not within target time | 28101 (22.1) | 21226 (75.5) | 6875 (24.5) | |
| Missing | 8083 (6.4) | 6314 (78.1) | 1769 (21.9) | |
| Procedure type | Internal fixation | 61888 (48.8) | 48735 (78.7) | 13153 (21.3) |
| Hemiarthroplasty | 54942 (43.3) | 42128 (76.7) | 12814 (23.3) | |
| Total Hip replacement | 9514 (7.5) | 8416 (88.5) | 1098 (11.5) | |
| Missing/Other | 553 (0.4) | 388 (70.2) | 165 (29.8) | |
| Calendar year of surgery | 2014 | 33361 (26.3) | 25853 (77.5) | 7508 (22.5) |
| 2015 | 50875 (40.1) | 40316 (79.2) | 10559 (20.8) | |
| 2016 | 42661 (33.6) | 33498 (78.5) | 9163 (21.5) | |
| Weekday of admission | Weekday | 84563 (66.6) | 66440 (78.6) | 18123 (21.4) |
| Weekend | 37966 (29.9) | 30269 (79.7) | 7697 (20.3) | |
| Missing | 4368 (3.4) | 2958 (67.7) | 1410 (32.3) | |
| Hospital volume | High volume | 65307 (51.5) | 51087 (78.2) | 14220 (21.8) |
| Medium volume | 29171 (23.0) | 23447 (80.4) | 5724 (19.6) | |
| Low volume | 32419 (25.6) | 25133 (77.5) | 7286 (22.5) | |
| ASA grade | I | 2814 (2.2) | 2559 (90.9) | 255 (9.1) |
| II | 33975 (26.6) | 29125 (85.7) | 4850 (14.3) | |
| III | 70338 (55.4) | 54474 (77.4) | 15864 (22.6) | |
| IV | 16440 (13.0) | 10932 (66.5) | 5508 (33.5) | |
| V | 284 (0.2) | 163 (57.4) | 121 (42.6) | |
| Missing | 3046 (2.4) | 2414 (79.3) | 632 (20.7) | |
| heart failure or pulmonary oedema | No | 107890 (85.0) | 85761 (79.5) | 22129 (20.5) |
| Yes | 13543 (10.7) | 9500 (70.1) | 4043 (29.9) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| chronic obstructive pulmonary diseases | No | 103301 (81.4) | 81612 (79.0) | 21689 (21.0) |
| Yes | 18132 (14.3) | 13649 (75.3) | 4483 (24.7) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| ischemic heart disease (acute) | No | 109335 (86.2) | 86249 (78.9) | 23086 (21.1) |
| Yes | 12098 (9.5) | 9012 (74.5) | 3086 (25.5) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| cardiac dysrhythmias | No | 93529 (73.7) | 74518 (79.7) | 19011 (20.3) |
| Yes | 27904 (22.0) | 20743 (74.3) | 7161 (25.7) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| ischemic heart disease (chronic) | No | 100351 (79.1) | 79520 (79.2) | 20831 (20.8) |
| Yes | 21082 (16.6) | 15741 (74.7) | 5341 (25.3) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| hypertension | No | 53007 (41.8) | 41381 (78.1) | 11626 (21.9) |
| Yes | 68426 (53.9) | 53880 (78.7) | 14546 (21.3) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| hypotension | No | 110695 (87.2) | 87580 (79.1) | 23115 (20.9) |
| Yes | 10738 (8.5) | 7681 (71.5) | 3057 (28.5) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| diabetes with complication | No | 119685 (94.3) | 93971 (78.5) | 25714 (21.5) |
| Yes | 1748 (1.4) | 1290 (73.8) | 458 (26.2) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| Alzheimer’s or dementia | No | 85056 (67.0) | 69150 (81.3) | 15906 (18.7) |
| Yes | 36377 (28.7) | 26111 (71.8) | 10266 (28.2) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| depression | No | 111356 (87.8) | 87546 (78.6) | 23810 (21.4) |
| Yes | 10077 (7.9) | 7715 (76.6) | 2362 (23.4) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| delirium | No | 111035 (87.5) | 87820 (79.1) | 23215 (20.9) |
| Yes | 10398 (8.2) | 7441 (71.6) | 2957 (28.4) | |
| Missing | 5464 (4.3) | 4406 (80.6) | 1058 (19.4) | |
| Admission from location | Own home/sheltered housing | 100760 (79.4) | 81611 (81.0) | 19149 (19.0) |
| Nursing care/residential care | 23127 (18.2) | 16049 (69.4) | 7078 (30.6) | |
| Rehabilitation unit | 155 (0.1) | 113 (72.9) | 42 (27.1) | |
| Acute hospital/already in | 2835 (2.2) | 1878 (66.2) | 957 (33.8) | |
| hospital/this hospital site/other hospital site of this trust/other hospital trust | ||||
| Missing | 20 (0.0) | 16 (80.0) | 4 (20.0) |
More than half were admitted to high volume hospitals (52%), on weekdays (67%), and underwent surgery within the target time (72%) (Table 1).
30-day survival
Overall
Among 126,897 patients, 99,667 (79%) patients mobilised early and 119,939 (94%) survived to 30-days post admission. The weighted odds ratio of survival at 30-days was 1.92 (95% CI 1.80–2.05) among those who mobilised early when compared with those who mobilised late (Table 2). The weighted probability (%) of survival at 30-days post-admission were 95.9 (95% CI 95.7–96.0) and 92.4 (95% CI 92.0–92.8) respectively among those who first mobilised early and those who first mobilised late (Figure 1).
Table 2. Odds ratios of 30-day survival by timing of mobilisation among patients surgically treated for non-pathological first hip fracture overall and across subgroups defined by dementia.
| Mobilisation timing | Patients n (%) | Survivors n (%) | Unweighted Odds Ratio (95% CI) | Adjusted Odds Ratio (95% CI)* |
|---|---|---|---|---|
| Overall(N=I26,897) | ||||
| mobilised 2 days or more after surgery | 27,230 (21.5) | 24,283 (89) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 99,667 (78.5) | 95,656 (96) | 2.89 (2.75 to 3.04) | 1.92 (1.80 to 2.05) |
| Without dementia (N=85,056) | ||||
| mobilised 2 days or more after surgery | 15,906 (18.7) | 14,548 (91) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 69,150 (81.3) | 67,207 (97) | 3.23 (3.01 to 3.47) | 2.06 (1.88 to 2.26) |
| With dementia (N=36,377) | ||||
| mobilised 2 days or more after surgery | 10,266 (28.2) | 8,915 (87) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 26,111 (71.8) | 24,393 (93) | 2.15 (2.00 to 2.32) | 1.77 (1.61 to 1.95) |
CI = confidence interval,
+ Unweighted Odds ratios from the analysis of complete case analysis for day 30 survival outcome includes 126,897 patients for the overall analysis, and 121,433 in the analysis by dementia.
using propensity weighting score that were calculated using a model to predict the mobilisation timing from the confounders variables: age, sex, ethnicity, comorbidities, fracture type, fracture surgery type, surgery timing, fracture year calendar, deprivation, week day of admission, prefracture residence, ASA grade, prefracture mobility and hospital volume. 35285, 24,582 and 10,703 cases with missing data of at lease one of these confounder variables were excluded from the overall, without dementia group and with dementia group respectively.
Figure 1. Weighted probability of survival at 30 days from admission in relation to timing of mobilization, overall, and by dementia.
By dementia
Overall, 36,377 (29%) patients had dementia at the time of presentation, 85,056 (67%) patients did not have dementia, and 5,464 (4%) had missing data for dementia. In total, 26,111 (72%) patients with dementia and 69,150 (81%) without dementia mobilised early (Supplementary File 1, Table S1). A total of 33,308 (92%) patients with dementia and 81,755 (96%) without dementia survived to 30-days post-admission.
The weighted odds ratios of survival at 30-days post-admission were 1.77 (95% CI 1.61 – 1.95) for those with dementia and 2.06 (95% CI 1.88 – 2.26) for those without dementia among those who mobilised early when compared with those who mobilised late (Table 2). The weighted probability of survival (%) at 30-days post-admission for those with dementia and without dementia were 93.3 (95% CI 92.9 – 93.6) and 97.0 (95% CI 96.8 – 97.2) respectively among those who first mobilised early, and were 88.7 (95% CI 87.9 – 89.4) and 94.0 (95% CI 93.6 – 94.4) respectively among those who first mobilised late (Figure 1).
30-day ambulation recovery
Overall
In total, 33,273 (26%) patients had complete data for timing of mobilisation and ambulation recovery at 30-days (Supplementary File 1, Table S2). Among these, 24,492 (74%) patients had outdoor ambulation prefracture and 8,781 (26%) patients had indoor ambulation only prefracture. Among those with outdoor ambulation prefracture, 20,820 (85%) patients mobilised early and 2,275 (9%) recovered their prefracture ambulation by 30-days postadmission. Among those with indoor ambulation prefracture, 6,517 (74%) patients mobilised early and 6,960 (79%) recovered their prefracture ambulation by 30-days postadmission.
Among those with outdoor ambulation prefracture, the weighted odds ratio of ambulation recovery at 30-days post-admission was 1.25 (95% CI 1.03 - 1.51) among those who mobilised early when compared with those who mobilised late (Table 3). Among those with indoor ambulation only prefracture, the weighted odds ratio of ambulation recovery at 30-days was 1.53 (95% CI 1.32 – 1.78) among those who mobilised early when compared with those who mobilised late (Table 3). The weighted probability of ambulation recovery given outdoor and indoor ambulation prefracture were 9.7 (95% CI 9.2 – 10.2) and 81.2 (95% CI 80.0 – 82.4) respectively among those who first mobilised early, and were 7.9 (95% CI 6.6-9.2) and 73.8 (95% CI 71.3-76.2) respectively among those who first mobilised late (Figure 2a).
Table 3. Odds ratios of 30-day recovery by timing of mobilisation among patients surgically treated for non-pathological first hip fracture across subgroups defined by mobility prefracture and dementia.
| Mobilisation timing | Patients n (%) | patients who recovered n (%) | Unweighted Odds Ratio (95% CI) | Weighted Odds Ratio (95% CI) |
|---|---|---|---|---|
| Outdoor ambulation (n = 24,492) | ||||
| mobilised 2 days or more after surgery | 3,672 (15.0) | 263 (7) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 20,820 (85.0) | 2012 (10) | 1.39 (1.21 to 1.58) | 1.25 (1.03 to 1.51)* |
| Indoor ambulation only (n = 8,781) | ||||
| mobilised 2 days or more after surgery | 2,264 (25.8) | 1629 (72) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 6,517 (74.2) | 5331 (82) | 1.75 (1.57 to 1.96) | 1.53 (1.32 to 1.78)* |
| Without dementia & outdoor ambulation (n= 16,761) | ||||
| mobilised 2 days or more after surgery | 2,375 (14.2) | 188 (8) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 14,386 (85.8) | 1371 (10) | 1.23 (1.05, 1.44) | 1.19 (0.96, 1.47)** |
| Without dementia & indoor ambulation (n = 3,797) | ||||
| mobilised 2 days or more after surgery | 924 (24.3) | 717 (78) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 2,873 (75.7) | 2509 (87) | 1.99 (1.65, 2.41) | 1.73 (1.35, 2.20)** |
| With dementia & outdoor ambulation (n = 4,338) | ||||
| mobilised 2 days or more after surgery | 862 (19.9) | 45 (5) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 3,476 (80.1) | 316 (9) | 1.82 (1.32, 2.50) | 1.52 (1.03, 2.23)** |
| With dementia & indoor ambulation (n = 3,982) | ||||
| mobilised 2 days or more after surgery | 1,077 (27.0) | 711 (66) | 1.00 | 1.00 |
| mobilised on the day of or day after surgery | 2,905 (73.0) | 2207 (76) | 1.63 (1.40, 1.90) | 1.52 (1.26, 1.84)** |
CI = confidence interval,
using propensity weighting score that were calculated using a model to predict the mobilisation timing from the confounders variables: age, sex, ethnicity, comorbidities, fracture type, fracture surgery type, surgery timing, fracture year calendar, deprivation, week day of admission, prefracture residence, ASA grade, prefracture ambulation and hospital volume. 12,947 cases with missing data of at lease one of these confounder variables were excluded from the main analysis. In the analysis by prefracture ambulation 9,650 and 3,297 with missing data were excluded from outdoor ambulation group and indoor ambulation group respectively, In the analysis by dementia condition 6,054 and 2,498 with missing data were excluded from the group without dementia and with dementia respectively.
using propensity weighting score that were calculated using a model to predict the mobilisation timing from the confounders variables: age, sex, ethnicity, comorbidities, fracture type, fracture surgery type, surgery timing, fracture year calendar, deprivation, week day of admission, prefracture residence, ASA grade, prefracture ambulation and hospital volume. 4,934, 1,120, 1,323, 1,175 cases with missing data of at lease one of these variables were excluded from “without dementia & outdoor ambulation”, “without dementia & indoor ambulation”, with dementia & outdoor ambulation”, with dementia & outdoor ambulation”, subgroups respectively.
Figure 2.
(a, b): a) Weighted probability of recovery at 30 days from admission in relation to timing of mobilization by ambulation prefracture. b) Weighted probability of recovery at 30 days from admission in relation to timing of mobilization by ambulation prefracture and dementia.
By dementia
Among the 33,273 patients with complete data for mobilisation timing and ambulation recovery at 30-days, 8,320 (25%) patients presented with dementia, 20,558 (62%) patients presented without dementia, and 4,395 (13%) had missing data for dementia. In total, 6,381 (77%) patients with dementia and 17,259 (84%) without dementia mobilised early (Supplementary File 1, Table S3). A total of 3,982 (48%) patients with dementia and 3,797 (19%) without dementia had indoor ambulation only prefracture. In total, 3,279 (39%) patients with dementia and 4,785 (23%) without dementia recovered their prefracture ambulation by 30-days post-admission.
Among those with dementia and outdoor ambulation prefracture, the weighted odds ratios of ambulation recovery at 30-days was 1.52 (95% CI 1.03 - 2.23) among those who mobilised early compared with those who mobilised late (Table 3). Among those without dementia and outdoor ambulation prefracture, the weighted odds ratios of ambulation recovery at 30-days was 1.19 (95% CI 0.96 - 1.47) among those who mobilised early compared with those who mobilised late (Table 3). The weighted probability of ambulation recovery (%) at 30-days among patients with dementia and outdoor ambulation only prefracture were 10.1 (8.9 – 11.3) and 6.9 (4.6 – 9.2) respectively for those who first mobilised early and for those who first mobilised late (Figure 2b). Among patients without dementia and outdoor ambulation only prefracture, these probabilities (%) were 9.6 (9.0 – 10.2) and 8.2 (6.7 – 9.7) respectively for those who first mobilised early and for those who first mobilised late.
Among those with dementia and indoor ambulation only prefracture, the weighted odds ratios of ambulation recovery at 30-days was 1.52 (95% CI 1.26 - 1.84) among those who mobilised early compared with those who mobilised late (Table 3). Among those without dementia and indoor ambulation prefracture, the weighted odds ratios of ambulation recovery at 30-days was 1.73 (95% CI 1.35 - 2.20) among those who mobilised early compared with those who mobilised late (Table 3). The weighted probability of ambulation recovery (%) at 30-days among patients with dementia and indoor ambulation only prefracture were 75.5 (73.7 -77.4) and 67.0 (63.4 -70.6) respectively for those who first mobilised early and for those who first mobilised late (Figure 2b). Among patients without dementia and indoor ambulation only prefracture, these probabilities (%) were 87.5 (86.0 -88.9) and 80.1 (76.9 -83.4) respectively for those who first mobilised early and for those who first mobilised late.
Sensitivity analyses
Full details of the potential influence of missing data in the exposure and potential confounders on the association between mobilisation timing and survival at 30-days are presented in Supplementary File 3. Results using the Within and Across approaches in these analyses were consistent and yielded similar estimates to those of the complete case analysis.
Discussion
Main findings
Early mobilisation was associated with 30-day survival and ambulation recovery after hip fracture surgery. The association between early mobilisation and 30-day ambulation recovery was stronger for those with indoor mobility only prefracture when compared to those with outdoor mobility prefracture. Patients with dementia were less likely to mobilise early despite similar associations for survival and ambulation recovery observed for those with and without dementia.
Comparison to previous literature
The findings of the current study support previous evidence of a beneficial association between mobilisation timing and outcomes after hip fracture surgery. This evidence reports reductions in complications,20 inhospital mortality,21 time to discharge from hospital,7 mortality at 6 months,5 return to home,22 ambulation at 30-days,6 and ambulation at 2 months.5 Here we provide additional evidence to support the survival benefits of early mobilisation from a large dataset with weighting for confounders. We build upon analyses of recovery by considering the outcome in relation to prefracture ambulatory ability. A notable new finding is by 30 days post-admission, only 9% of those with outdoor ambulation prefracture recovered this ability whilst 80% of those with indoor ambulation only recovered (among those mobilised early).
Interpretation
We report early mobilisation was associated with survival and ambulatory recovery at 30-days. This benefit is in keeping with other established indicators of care quality after hip fracture. For example, early surgery (on the day of or day after admission) was associated with survival at 30-days in a cohort of 139,119 patients in Canada,23 and admission to an orthogeriatric unit (vs. orthopaedic unit) was associated with survival in a cohort of 11,461 patients in Denmark.24 However, early mobilisation is not a standard performance indicator in audits of hip fracture care.25 Where captured, the proportion of patients mobilised early ranges from 55% in New Zealand to 90% in Denmark.25 We propose early mobilisation as a measured indicator of performance internationally. This would enable clinicians to determine the extent to which they are achieving early mobilisation with their patients and to evaluate the benefit of quality improvement initiatives to improve performance.
Patients with dementia benefitted in terms of survival and ambulation recovery from early mobilisation. Yet, fewer patients with dementia (72%) mobilised early compared to those without dementia (81%). This suggests practice variation may deny optimal recovery chances for patients with and without dementia. Physiotherapists’ recently reported pressure to adhere to guidelines ‘which may not be achievable or appropriate for those with dementia’.26 A better understanding of appropriate management after hip fracture for patients with dementia and subsequent therapist training may be warranted.
Limitations
There is potential for bias due to residual confounding by variables associated with early mobilisation and survival/recovery including those related to the patient (e.g., motivation, frailty, acute illness), operation (e.g., intra-operative fracture, wound haematoma, anaesthetic type), admission (e.g. weekend) or overall standard of hip fracture care (e.g. hospitals with understaffed therapy services may also be deficient in other aspects of hip fracture care). This may have led to an overestimation of the association between early mobilisation and outcomes reported here. There is potential for bias due to data quality27 and/or data missingness. We did not impute missing data for our analysis of 30-day recovery as data was not missing at random.15 We reported similar findings from imputed and complete case analysis of 30-day survival. There is potential for misclassification bias where clinicians entering data may interpret the ‘ability to sit or stand out of bed’ as without hoist transfer. This may lead to an underestimation of the association between mobilisation timing and 30-day outcomes. We classified patients as ‘with dementia’ or ‘without dementia’ as we did not have information related to the disease stage. Patients with new-onset dementia may have a different chance of early mobilisation, survival and/or recovery, when compared to those with advanced/end-stage dementia. Further, pre-existing cognitive impairment is associated with delirium and further cognitive decline following surgery.28 We did not capture this potential change in cognitive status in our classification by dementia. Lastly our data is not generalizable to populations with different patient demographics receiving care along different pathways.
Future Research
The average figures reported here do not reveal the considerable variation in rates of early mobilisation reported by different hospitals in the UK. In 2015, the median percentage of patients mobilised early was 77%, (Q1-Q3: 61.9% – 90.6%), far more than can be explained by differences in case-mix.10 In 2017, a UK national audit of physiotherapy after hip fracture highlighted staffing, lack of equipment, pain control, hypotension, agitation/refusal, and poor prefracture function as barriers to early mobilisation.29 There is a need to quantify the extent to which these parameters moderate the association between mobilisation timing and outcomes. This would determine which aspects of care to target for improvement and ultimately patient benefit.
The differences in ambulation recovery by 30-days in relation to prefracture ambulatory ability has significance for service delivery and clinical practice. In 2015, the average length of stay (acute and post-acute inpatient NHS facility) was 21 days with only two hospitals reporting average stays above 30 days.10 Rehabilitation beyond 30-days is required to support recovery, particularly for those with better prefracture ambulation. Arrangements for ongoing community-based rehabilitation are highly variable across the UK30 and are estimated by national audit of intermediate care to be insufficient to meet the clinical need.31 Moreover, whether outdoor ambulation is consistently incorporated into community-based rehabilitation after hip fracture is uncertain.
Conclusion
Early mobilisation (ability to sit or stand out of bed on the day of or day after surgery) was associated with survival and recovery for patients with and without dementia after hip fracture. Early mobilisation should be incorporated as a measured indicator of quality internationally. Reasons for failure to mobilise early should be captured to inform quality improvement initiatives.
Supplementary Material
Acknowledgements
We are grateful to NHS Digital, NHS Wales Informatics Service, and the Royal College of Physician’s Falls and Fragility Fracture Audit programme for providing the data used in this study. The views expressed in this publication are those of the authors and do not necessarily reflect those of the NHS or the Department of Health and Social Care. This publication is based on data collected by or on behalf of Healthcare Quality Improvement Partnership, who have no responsibility or liability for the accuracy, currency, reliability and/or correctness of this publication.
Funding source
This paper presents independent research funded by the National Institute for Health Research (NIHR) under its Research for Patient Benefit (RfPB) Programme (Grant Reference Number PB- PG-1216-20031). The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.
References
- 1.Royal College of Physicians. National Hip Fracture Database annual report 2019. RCP; London: [Google Scholar]
- 2.Neuburger J, Currie C, Wakeman R, et al. The Impact of a National Clinician-led Audit Initiative on Care and Mortality after Hip Fracture in England: An External Evaluation using Time Trends in Non-audit Data. MedCare. 2015;53(8):686–91. doi: 10.1097/MLR.0000000000000383. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.National Clinical Guideline Centre. The management of hip fracture in adults. National Clinical Guidelines Centre; London: 2011. [Google Scholar]
- 4.Benchimol EI, Smeeth L, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med. 2015;12(10):e1001885. doi: 10.1371/journal.pmed.1001885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Siu AL, Penrod JD, Boockvar KS, et al. Early ambulation after hip fracture: effects on function and mortality. ArchInternMed. 2006;166(7):766–71. doi: 10.1001/archinte.166.7.766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Su B, Newson R, Soljak H, et al. Associations between post-operative rehabilitation of hip fracture and outcomes: national database analysis. BMC Musculoskelet Disord. 2018;19(1):211. doi: 10.1186/s12891-018-2093-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sheehan KJ, Goubar A, Almilaji O, et al. Discharge after hip fracture surgery by mobilisation timing: secondary analysis of the UK National Hip Fracture Database. Age Ageing. 2020 doi: 10.1093/ageing/afaa204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Bai J, Zhang P, Liang X, et al. Association between dementia and mortality in the elderly patients undergoing hip fracture surgery: a meta-analysis. J Orthop Surg Res. 2018;13(1):298. doi: 10.1186/s13018-018-0988-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Huusko TM, Karppi P, Avikainen V, et al. Randomised, clinically controlled trial of intensive geriatric rehabilitation in patients with hip fracture: subgroup analysis of patients with dementia. BMJ. 2000;321(7269):1107–11. doi: 10.1136/bmj.321.7269.1107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Royal College of Physicians. National Hip Fracture Database annual report 2016. RCP; London: [Google Scholar]
- 11.StataCorp. Stata Statistical Software: Release 16. StataCorp LLC; College Station, TX: 2019. [Google Scholar]
- 12.Leuven E, Sianesi B. PSMATCH2: Stata Module to Perform Full Mahalanobis and Propensity Score Matching, Common Support Graphing, and Covariate Imbalance Testing [program] 2003
- 13.Newson Roger. SCENTTEST: Stata module to compute scenario arithmetic means and their difference. 2015 https://EconPapers.repec.org/RePEc:boc:bocode:s457904.
- 14.Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res. 2011;46(3):399–424. doi: 10.1080/00273171.2011.568786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.White IR, Royston P, Wood AM. Multiple imputation using chained equations: Issues and guidance for practice. Stat Med. 2011;30(4):377–99. doi: 10.1002/sim.4067. [DOI] [PubMed] [Google Scholar]
- 16.Graham JW, Olchowski AE, Gilreath TD. How many imputations are really needed? Some practical clarifications of multiple imputation theory. Prev Sci. 2007;8(3):206–13. doi: 10.1007/s11121-007-0070-9. [DOI] [PubMed] [Google Scholar]
- 17.Rubin DB. Multiple imputation for nonresponse in surveys. John Wiley and Sons; New York: 1987. [Google Scholar]
- 18.Penning de Vries B, Groenwold R. Comments on propensity score matching following multiple imputation. Stat Methods Med Res. 2016;25(6):3066–68. doi: 10.1177/0962280216674296. [DOI] [PubMed] [Google Scholar]
- 19.Mitra R, Reiter JP. A comparison of two methods of estimating propensity scores after multiple imputation. Stat Methods Med Res. 2016;25(1):188–204. doi: 10.1177/0962280212445945. [DOI] [PubMed] [Google Scholar]
- 20.Kamel HK, Iqbal MA, Mogallapu R, et al. Time to ambulation after hip fracture surgery: relation to hospitalization outcomes. J Gerontol A Biol Sci Med Sci. 2003;58(11):1042–5. doi: 10.1093/gerona/58.11.m1042. [DOI] [PubMed] [Google Scholar]
- 21.Ferris H, Brent L, Coughlan T. Early mobilisation reduces the risk of in-hospital mortality following hip fracture. Eur Geriatr Med. 2020 doi: 10.1007/s41999-020-00317-y. [DOI] [PubMed] [Google Scholar]
- 22.Oldmeadow LB, Edwards ER, Kimmel LA, et al. No rest for the wounded: early ambulation after hip surgery accelerates recovery. ANZ J Surg. 2006;76(7):607–11. doi: 10.1111/j.1445-2197.2006.03786.x. [DOI] [PubMed] [Google Scholar]
- 23.Sobolev B, Guy P, Sheehan KJ, et al. Mortality effects of timing alternatives for hip fracture surgery. CMAJ. 2018;190(31):E923–E32. doi: 10.1503/cmaj.171512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kristensen PK, Thillemann TM, Soballe K, et al. Can improved quality of care explain the success of orthogeriatric units? A population-based cohort study. Age Ageing. 2016;45(1):66–71. doi: 10.1093/ageing/afv155. [DOI] [PubMed] [Google Scholar]
- 25.Johansen A, Golding D, Brent L, et al. Using national hip fracture registries and audit databases to develop an international perspective. Injury. 2017;48(10):2174–79. doi: 10.1016/j.injury.2017.08.001. [DOI] [PubMed] [Google Scholar]
- 26.Hall AJ, Watkins R, Lang IA, et al. The experiences of physiotherapists treating people with dementia who fracture their hip. BMC Geriatr. 2017;17(1):91. doi: 10.1186/s12877-017-0474-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Cundall-Curry DJ, Lawrence JE, Fountain DM, et al. Data errors in the National Hip Fracture Database: a local validation study. Bone Joint J. 2016;98-B(10):1406–09. doi: 10.1302/0301-620X.98B10.37089. [DOI] [PubMed] [Google Scholar]
- 28.Uzoigwe CE, O’Leary L, Nduka J, et al. Factors associated with delirium and cognitive decline following hip fracture surgery. Bone Joint J. 2020;102-B(12):1675–81. doi: 10.1302/0301-620X.102B12.BJJ-2019-1537.R3. [DOI] [PubMed] [Google Scholar]
- 29.Royal College of Physicians. Recovering after a hip fracture: helping people understand physiotherapy in the NHS. Physiotherapy ‘Hip Sprint’ audit report; London: 2017. Falls and Fragility Fracture Audit Programme. [Google Scholar]
- 30.Neuburger J, Harding KA, Bradley RJ, et al. Variation in access to community rehabilitation services and length of stay in hospital following a hip fracture: a cross-sectional study. BMJ Open. 2014;4(9):e005469. doi: 10.1136/bmjopen-2014-005469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.National Benchmarking Network. National Audit of Intermediate Care. Summary Report - England. Assessing progress in services aimed at maximising independence and reducing use of hospitals. NHS England; 2017. pp. 38–41. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.