Complications following hip fracture: Results from the World Hip Trauma Evaluation cohort study.

INTRODUCTION: Hip fractures are common in people over 60 years of age, and are associated with significant disability, morbidity and mortality. The aim of this study was to record the incidence of complications in the first 120 days following hip fracture.
METHODS: The World Hip Trauma Evaluation (WHiTE) study is a multicentre, prospective cohort study conducted in National Health Service (NHS) hospitals in England and Wales. Participants are 60 years and older who received operative treatment for a hip fracture. We report the incidence of complications recorded by hospital staff until discharge from hospital and by participants at 120-days post-surgery.
RESULTS: An analysis of 8673 consecutive participants enrolled in the WHiTE study revealed the following risks of complications within the first 120 days: signs of wound infection (3.1%); dislocation (0.5%); failure of fixation (0.6%); peri‑prosthetic fracture (0.3%); overall revision surgery (0.9%); blood loss requiring transfusion (6.1%); chest infection (6.3%); urinary tract infection (5.0%); deep vein thrombosis/pulmonary embolus (1.8%); cerebrovascular accident (0.6%); acute coronary syndrome/myocardial infarction (0.6%); acute kidney injury (1.3%).
CONCLUSION: The rates of complications reported here provide a reference range against which future studies might be assessed. Registration: ISRCTN63982700.

Introduction

Background

Hip fractures are common in people over 60 years of age, and are associated with significant disability, morbidity and mortality [1], [2], [3], [4]. These patients are susceptible to developing a variety of post-operative complications related to the treatment received or hospital stay [5]. These include complications directly related to the surgery such as wound infection, dislocation, failure of fixation, peri‑prosthetic fracture and blood transfusion, as well as other complications such as chest infection, urinary tract infection (UTI), venous thromboembolism (VTE), cerebrovascular accident (CVA), myocardial infarction (MI), and acute kidney injury (AKI). Post-operative complications hinder rehabilitation, increase length of hospitalization and result in excess morbidity and mortality in an already high-risk patient group [5], [6].

The incidence of these complications in hip fracture populations have been reported in other cohort studies and is summarised in Table 1 [4], [7], [8], [9], [10], [11], [12], [13], [14]. However, there remains a paucity of data available regarding complication risks following contemporary treatments. Furthermore, no studies have recorded patient-reported complications following discharge from hospital, instead relying on information obtained from medical records. The effective management of patients with hip fracture requires an understanding of the post-operative complications to enable optimization of care.

Table 1

Reported rates (%) of complications in other cohort studies. All rates are given as a percent of all patients, "-" indicates complications that were not reported by that study.

	Prieto-Alhambra et al. [7]	Hansson et al. [8]	Inacio et al. [9]	McNamara et al. [10]	Gjertsen et al. [11]	Rosencher et al. [12]	Roche et al. [13]	Lawrence et al. [14]	Griffin et al. [4]
Sample size	997	664	12,562	5300	13,251	6860	2448	8930	741
Complications reporting period	Inpatient	1 year	1 year	3 months	3 years	Inpatient (VTE 3 months)	Inpatient	Inpatient	1 year
Design	Prospective	Retrospective	Retrospective	Prospective	Prospective	Prospective	Prospective	Retrospective	Prospective
Chest infection	8	11	11.4	–	–	2.7	9	–	4.1
Urinary tract infection	9.7	–	–	–	–	10.3	4	–	3.3
Cerebrovascular accident	–	2	–	–	–	0.7	1	1	0.1
Acute coronary syndrome/MI	–	2	0.9	–	–	0.5	1	–	0
Blood transfusion	–	–	–	–	–	38.2	–	–	0.5
Venous thromboembolism	–	3	2.6	2.2	–	1.3	2	1	1.1
Dislocation	–	0.5	0.9	–	0.4	–	–	–	1.8
Failure of fixation	–	–	–	–	4.5	–	–	–	1.6
Wound infection	0.8	4	1.1	–	–	–	–	–	2.1
Acute kidney injury	14.1	–	–	–	–	–	–	0.09	–
Peri-implant fracture	–	1	–	–	0.5	–	–	–	–
Non-union	–	2	–	–	1.7	–	–	–	–
Revision surgery	–	7	2.4	–	17.5	–	–	–	–

Here, we report the incidence of complications in the World Hip Trauma Evaluation (WHiTE) cohort [15]. Assessments of the WHiTE cohort have been published previously [16], [17], [18], [19], [20].

Methods

Design, setting & participants

The WHiTE cohort is a prospective observational cohort study that collects information on assessment, treatment and recovery of patients admitted to participating United Kingdom National Health Service (NHS) hospitals with a hip fracture. Patients were eligible to participate in the WHiTE cohort study if they were 60 years or older and were to be treated operatively for a hip fracture. Some patients were recruited via their participation in the embedded randomised controlled trials, WHiTE3 Hemi, WHiTE4 or WHiTE5 (in progress) [21], [22].

Consent

Consent to participate in the WHiTE cohort study was granted by patients or an agreement was provided by an appropriate consultee for patients without capacity. Participants were prospectively consented for their involvement in the study. The WHiTE studies were approved by research ethics committees (REC; WHiTE cohort approved by Camberwell St Giles Research Ethics Committee with reference 11/LO/0927, WHiTE3 Hemi and WHiTE4 by the West Midlands Coventry and Warwickshire REC – 14/WM/1098 and 16/WM/0001 respectively, and WHiTE5 by Wales REC 5 – 16/WA/0351).

Treatment

Participants enrolled in the WHiTE cohort study were treated in accordance with local standard care pathways. The National Institute for Clinical Excellence (NICE) has issued standardised care guidelines that are used in the majority of hospitals and are summarised in the WHiTE cohort study protocol [15]. A minority of participants were enrolled in embedded randomised studies and were randomised to a treatment, all of which were in routine use in the NHS [21], [22].

Sample size & dataset

A pre-specified analysis of the WHiTE cohort was planned to include the first 6000 complete outcome sets as per the published protocol [15].

Data collection

A list of complications known to be related to the surgery or inpatient stay were defined from previously reported data [4]. The complications directly related to the surgery were specified to be: dislocation, failure of fixation, wound infection, peri‑prosthetic fracture, non-union, and revision surgery (all-cause). Other complications were chest infection, UTI, CVA, acute coronary syndrome and MI, blood transfusion, AKI, and VTE.

Hospital research teams were asked to report a record of diagnoses of complications from the medical record using clinical reporting forms completed at baseline/discharge. At 120 days post-surgery, participants or their carer completed a follow-up questionnaire administered over the phone or by post. They were asked to report if they experienced any signs of wound infection, dislocation, failure of fixation, peri‑prosthetic fracture, blood transfusion, chest infection, UTI, VTE, CVA, MI or AKI during the 120-day period post-surgery. Signs of wound infection were purulent drainage from the wound, prescription of antibiotics for the wound, or reoperation for the purpose of washout or debridement of the wound.

A minority of patients chose to enroll in the study but opted out of 120-day follow-up post-surgery, and instead provided only routine hospital data. In these patients, death, cause of death, inpatient complications and serious adverse events (SAEs) were recorded but no patient-reported complications were collected.

Data sources and coding

Data were entered into the WHiTE database (OpenClinica V3.7, OpenClinica LLC, Waltham, MA, USA) directly by the central trial team for questionnaires and forms completed by telephone, or transcribed from paper questionnaires completed by site staff or participants. Free text details of complications were manually coded against the pre-specified list of complications.

Results

At the time of the pre-planned data-lock, the WHiTE cohort contained data from 8673 participants recruited across 20 sites [15]. The participants and interventions in the WHiTE study were representative of those reported in the National Hip Fracture Database (NHFD), and have been described in detail by Metcalfe et al. [17].

Of the 8673 participants, 1060 (12%) died before 120 days. Of the remaining 7613 participants, 6100 (80%) were contacted by telephone or post at the 120-day follow-up post-surgery. The remaining participants opted out of a 120-day follow-up, withdrew from the study or could not be contacted.

The incidences of each related complication reported up to 120 days are given in Table 2. Estimated trends in cumulative rates (post-surgery) for each of the complications in Table 2, are shown in Fig. 1. The figures show that blood transfusions occurred relatively quickly, 50% of events were within three days of surgery; whereas chest infections and UTIs generally occurred later, 50% of events were within six and five days of surgery respectively.

Table 2

Reported risks of complications in the WHiTE study. Percentage shown is of all 8673 patients enrolled.

Complication related to surgery	Number of patients	% of patients	95% Confidence interval
Blood transfusion	527	6.08%	5.59 – 6.60%
Any sign of wound infection	271	3.12%	2.77 – 3.52%
Revision surgery (all cause)	75	0.86%	0.69 – 1.09%
Failure of fixation	48	0.55%	0.41 – 0.74%
Dislocation	46	0.53%	0.39 – 0.71%
Peri-prosthetic fracture	29	0.33%	0.23 – 0.49%
Other complications
Chest infection	550	6.34%	5.84 – 6.88%
UTI	431	4.97%	4.53 – 5.45%
DVT/PE	159	1.83%	1.57 – 2.14%
Acute kidney injury	113	1.30%	1.08 – 1.57%
Myocardial infarction/acute coronary	55	0.63%	0.48 – 0.83%
Cerebrovascular accident	52	0.60%	0.45 – 0.79%

Fig. 1

Estimated complication rates in the first 120 days post-surgery for (a) complications directly related to surgery and (b) other complication.

Of the complications reported by patients at 120 days, 278 of 713 (39%) were directly related to the surgery, 88 (12%) were other complications, and the remainder were unrelated to the injury or treatment. Of the 366 patient reported complications related to the injury, 282 (77%) had not already been reported by hospital research teams. Signs of wound infection were the most frequently reported complication by patients at 120 days. The frequency of patient-reported wound infection and site-reported diagnoses of wound infection, are summarised in Table 3.

Table 3

Reported risks of wound infection in the WHiTE study. Percentage shown is of all 8673 patients enrolled. Some wound infections were reported by both the patient and the hospital. Patients reported multiple signs of infection.

	Number of patients	% of all patients
Any diagnosis or sign of wound infection	271	3.1%
Wound infection reported by hospital staff	110	1.3%
Patient reported signs of wound infection:	202	2.3%
Purulent drainage >5 days	44	0.5%
Antibiotics for wound	155	1.8%
Surgical debridement/washout	28	0.3%
Microbiological confirmation of infection	115	1.3%

Discussion

The WHiTE cohort is the largest prospective cohort study of hip fracture outcomes in the world. The high rates of 120-day follow-up of participants allows for collection of complications even after discharge from hospital.

Chest infection was the most frequently occurring complication in this cohort at 6.3%, which is comparable with from other hip fracture cohorts (Table 1). However, the incidence of chest infection in our cohort was higher than the 5.3% reported in a meta-analysis of randomised controlled trials (RCTs) of patients undergoing surgery for hip fractures [23]. It is possible that our approach to data collection, which combined patient-reporting with information collected from medical records results in a higher detection rate for some post-operative complications. Additionally, the RCTs included in the meta-analysis only reported complications occurring in the immediate post-operative period. This is in contrast with our cohort, where all complications occurring up to 120 days were included. Since many RCTs exclude patients with cognitive impairment, it is also possible that fitter participants are selected into trials compared to the observational WHiTE cohort, which is representative of all patients with hip fracture in the UK [17].

Urinary tract infection occurred in 5.0% of our cohort, which is similar to the study by Roche et al. [13]. Comparatively, the incidence of urinary tract infection was 9.7% and 10.3% of patients in the Spanish Registry of Femur Fractures and ESCORTE study respectively [7], [12]. This discrepancy may be due to international variations in clinical practice; specifically with regards to the prescription of antibiotics and the use of urinary catheters. In the United Kingdom, urinary catheterisation during the peri‑operative period is typically avoided except in specific circumstances. Early removal and trial without catheter is advocated in patients who require catheterisation peri‑operatively. The comparable risk of UTI between the present study and that by Roche et al., which was also conducted in the UK may reflect adherence to national guidelines.

There appears to be a discrepancy in the reporting of wound infection rates depending on the source of data. The risk of wound infection as reported by patients is higher than that reported by hospital staff at 2.3% compared to 1.3%. This could be due to several factors including the lack of well-defined diagnostic criteria for wound infections, prophylactic antibiotic use and recall bias. Nonetheless, other studies have reported risks of wound infection that range between 0.8% to 4.0%, which are consistent with our findings [4], [7], [8], [9].

Revision surgery was performed in a small proportion of patients (0.9%) in our cohort. This was due to failure of fixation in 60% and peri‑prosthetic fracture in 37.3%. Further surgery for ‘non-union’ was very rare (2.7%), but there may well be some overlap with those patients having further surgery for failure of fixation. Both the Kaiser Permanante Hip Fracture Registry and Irish Hip Fracture Database report similar figures at 2.5% lifetime reoperation rate and 1.1% reoperation rate at 30 days post-surgery respectively [9], [24]. This is markedly lower than the re-operation rates reported in the Danish Fracture Database at 10.9% over two years and lower than the number of reoperations reported in the Norwegian Hip Fracture Registry (total of 2325 reoperations and 13,251 primary operations during 2015 and 2016), which is because both databases track all subsequent re-operations and not just revisions [11], [25]. Furthermore, the choice of primary procedure may contribute to the differences observed – screw osteosynthesis was the primary procedure performed in a large proportion of intracapsular fractures in both the Norwegian and Danish cohorts, which is in contrast to our cohort where hemiarthroplasties were more frequently performed.

Mortality in our patient cohort at 120 days was 12.2%, which is comparable with data from the Spanish Registry of Femur Fractures and Norwegian Hip Fracture Register at 10.2% and 14.0% respectively [7], [11].

Overall, the risk of complications reported here is higher than in other studies. The combination of medical records review and patient-report at 120 days used in the WHiTE cohort may provide a more accurate estimate than reviews of inpatients medical records alone, and provides a reference range against which future studies can be compared.

Strengths and limitations

The WHiTE study includes a variety of hospitals in England and Wales and combines patient-reported and hospital-reported outcomes up to four months post-surgery. Over 83% of patients enrolled in the study provided outcome data (completed questionnaire or died) at 120 days post-surgery.

Reporting by sites and patients did not always agree. In some instances patients reported a complication but no record was found at the hospital. In other instances, sites notified the trial office of complications that had not been reported by the patients. Some patients may have been treated at other hospitals, in which case the medical record could not be accessed, but this is rare in the UK where the great majority of patients with hip fracture are treated in their local hospital. This discrepancy is likely to reflect problems with recall in this frail group of patients but equally may reflect poor record keeping in some hospitals. The recording of wound healing complications, including infection, was poor in many cases.

Another limitation is the overlap between indications for the prescription of antibiotics; most notably for chest, wound or urinary tract infection. Although most patients were able to recall that they were prescribed antibiotics, they were not always certain about the indication for the prescription. However, in the absence of a single medical record covering hospital and community treatments, the combination of hospital record review and patient report at 120 days is likely to represent the most accurate data available in the UK to date.

For a number of events for each complication, it was impossible to locate the exact date of the complication, due mainly to insufficient data provided by the study participants, so the events were assumed, for the purposes of exposition to have occurred at 120 days. This is apparent as a marked upward jump in rates at precisely 120 days for a number of the complications (e.g. wound infections and chest infections).

Conclusion

The present study provides prospectively collected data on the incidence of complications up till four months following surgery for hip fracture in the UK. The rates of complications reported here therefore provide a reference range against which future studies may be assessed.

Funding

The WHiTE study is funded by the NIHR Oxford Biomedical Research Centre. WHiTE4 is funded by X-BOLT Ltd. WHiTE5 is funded by the Research for Patient Benefit (PB-PG-0215–36,043 and PB-PG-1216–20,021). XG is funded by the Clinician Scientist fellowship award. MC is an NIHR Senior Investigator.

WHiTE, and each of the embedded trials, is supported by the Oxford Biomedical Research centre. The views expressed here are those of the authors and not the NIHR, Department of Health, Oxford BRC or X-BOLT.

WHiTE collaborators

Mehool Acharya (PI; North Bristol NHS Trust)

Paul Baker (PI; South Tees Hospitals NHS Foundation Trust)

Callum Clark (PI; Frimley Health NHS Foundation Trust)

Andrew Clarke (Trainee PI; South Tees Hospitals NHS Foundation Trust)

Avi Das (Trainee PI; Nottingham University Hospitals NHS Trust)

John Davison (PI; University Hospitals of Leicester NHS Trust)

Paul Dixon (PI; City Hospitals Sunderland NHS Foundation Trust)

Mark Dunbar (PI; University Hospitals Birmingham NHS Foundation Trust)

William Eardley (PI; South Tees Hospitals NHS Foundation Trust)

Mark Farrar (PI; Poole Hospital NHS Foundation Trust)

Paul Fearon (PI; Newcastle upon Tyne Hospitals NHS Foundation Trust)

Ad Gandhe (PI; Portsmouth Hospitals NHS Trust)

Bob Handley (PI; Oxford University Hospitals NHS Foundation Trust)

Peter Hull (PI; Cambridge University Hospitals NHS Foundation Trust)

Andrew McAndrew (PI; Royal Berkshire NHS Foundation Trust)

Damian McClelland (PI; University Hospitals of North Midlands NHS Trust)

Iain McNamara (PI; Norfolk and Norwich University Hospitals NHS Foundation Trust)

Ben Ollivere (PI; Nottingham University Hospitals NHS Trust)

Mike Reed (PI; Northumbria Healthcare NHS Foundation Trust)

Benedict Rogers (PI; Brighton and Sussex University Hospitals NHS Trust)

Graham Smith (PI; Frimley Health NHS Foundation Trust)

Jonathan Young (PI; University Hospitals Coventry and Warwickshire NHS Trust)

Declaration of Competing Interest

WHiTE4 is funded by X-BOLT Ltd. The views expressed here are those of the authors and not X-BOLT. Although none of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article, benefits that have been or will be received will be directed solely to a research fund, foundation, educational institution, or other non-profit organization with which one or more of the authors are associated.