IMPACT-Global Hip Fracture Audit: Nosocomial infection, risk prediction and prognostication, minimum reporting standards and global collaborative audit: Lessons from an international multicentre study of 7,090 patients conducted in 14 nations during the COVID-19 pandemic.

AIMS: This international study aimed to assess: 1) the prevalence of preoperative and postoperative COVID-19 among patients with hip fracture, 2) the effect on 30-day mortality, and 3) clinical factors associated with the infection and with mortality in COVID-19-positive patients.
METHODS: A multicentre collaboration among 112 centres in 14 countries collected data on all patients presenting with a hip fracture between 1st March-31st May 2020. Demographics, residence, place of injury, presentation blood tests, Nottingham Hip Fracture Score, time to surgery, management, ASA grade, length of stay, COVID-19 and 30-day mortality status were recorded.
RESULTS: A total of 7090 patients were included, with a mean age of 82.2 (range 50-104) years and 4959 (69.9%) being female. Of 651 (9.2%) patients diagnosed with COVID-19, 225 (34.6%) were positive at presentation and 426 (65.4%) were positive postoperatively. Positive COVID-19 status was independently associated with male sex (odds ratio (OR) 1.38, p = 0.001), residential care (OR 2.15, p < 0.001), inpatient fall (OR 2.23, p = 0.003), cancer (OR 0.63, p = 0.009), ASA grades 4 (OR 1.59, p = 0.008) or 5 (OR 8.28, p < 0.001), and longer admission (OR 1.06 for each increasing day, p < 0.001). Patients with COVID-19 at any time had a significantly lower chance of 30-day survival versus those without COVID-19 (72.7% versus 92.6%, p < 0.001). COVID-19 was independently associated with an increased 30-day mortality risk (hazard ratio (HR) 2.83, p < 0.001). Increasing age (HR 1.03, p = 0.028), male sex (HR 2.35, p < 0.001), renal disease (HR 1.53, p = 0.017), and pulmonary disease (HR 1.45, p = 0.039) were independently associated with a higher 30-day mortality risk in patients with COVID-19 when adjusting for confounders.
CONCLUSION: The prevalence of COVID-19 in hip fracture patients during the first wave of the pandemic was 9%, and was independently associated with a three-fold increased 30-day mortality risk. Among COVID-19-positive patients, those who were older, male, with renal or pulmonary disease had a significantly higher 30-day mortality risk.

Introduction

The coronavirus disease 2019 (COVID-19) pandemic disrupted the delivery of Trauma and Orthopaedic (T&O) services, but despite a reduction in the incidence of activity-related trauma the incidence of fragility-related trauma was unchanged.[1], [2], [3] Developing COVID-19 in the perioperative period has been reported to double the background mortality risk following orthopaedic surgery, and the patients at greatest risk of mortality from COVID-19 are those who are older, comorbid and presenting with a fragility fracture. It is essential to have an understanding of the prevalence and patterns of SARS-CoV-2 infection within the hip fracture population, and to analyse the effects of the COVID-19 pandemic on this large and vulnerable patient group.

A recent systematic review and meta-analysis found that hip fracture patients with COVID-19 had a crude 30-day mortality of 35% and was seven times the risk of patients without COVID-19. However, in this same review less than half of the included studies reported patient age and sex and only two adjusted for confounding factors in their analysis. , Two multicentre cohort studies by the International Multicentre Project Auditing COVID-19 in Trauma & Orthopaedics in Scotland (IMPACT-Scot) Group have reported that after adjusting for confounding factors the 30-day mortality risk in COVID-19-positive hip fracture patients was three times greater than in COVID-19-negative patients. Furthermore, the reports are from a single nation with a relatively homogenous population and a standardised approach to hip fracture services.[4], [6], [7]

The IMPACT Global Hip Fracture Audit aimed to determine factors associated with a positive COVID-19 diagnosis and the influence this has on outcome, with the inclusion of international data from a wider range of patients and healthcare providers from across the globe. The aims of this international multicentre audit were to examine the hip fracture population and assess the: 1) prevalence and clinical factors associated with a diagnosis of COVID-19 in the preoperative and postoperative periods; 2) the independent effect of COVID-19 on 30-day mortality, and 3) factors associated with mortality in COVID-19-positive patients.

Patients and methods

In March 2020 the International Multicentre Project Auditing COVID-19 in Trauma & Orthopaedics (IMPACT) was established in order to provide an emergency clinical audit response to the COVID-19 pandemic. , It was recognised that investigation into the effects of COVID-19 on hip fracture patients and services was necessary and urgent. The IMPACT collaborative network gained support from the Scottish Hip Fracture Audit (SHFA), Scottish Government and the Scottish Committee for Orthopaedics & Trauma (SCOT). An international multicentre observational cohort study was subsequently established with data collected retrospectively from 112 hospitals in 14 nations, including: Australia, Argentina, Chile, Cyprus, England, India, Italy, Greece, Mexico, Northern Ireland, Scotland, Spain, Sudan, Wales. Centres were invited to participate through a recruitment process delivered through existing hospital networks and audit programmes, the Fragility Fracture Network (FFN) and the Royal College of Surgeons of England.

Data were collected in accordance with UK Caldicott guidance and equivalent principles in each nation, and no patient-identifiable information was transferred outside of local units or accessed by the IMPACT research team.

Inclusion and exclusion criteria

All patients who were over 50 years of age and presenting with a hip fracture to any participating hospital in the study period (1st March 2020 to 31st May 2020) were included. The inclusion criteria were that of the SHFA and previous IMPACT reports: all intracapsular or extracapsular fractures of the femur proximal to and including the distal limit of the subtrochanteric region (defined as a point five centimetres distal to the lesser trochanter). Periprosthetic femur fractures and isolated fractures of the pubic rami, acetabulum, and greater trochanter were excluded.

Baseline data collection

Data collection was defined prior to the commencement of the audit, which was delivered by a team of data collectors (comprised of clinicians and trained auditors) who were local to each hospital. Patients were identified through retrospective review of local admission data throughout the study period, and these data were cross-referenced with patients’ medical records, surgical operating lists and discharge letters. Data were input into the IMPACT Hip Fracture Audit data collection tool, a database constructed with data-validated fields and automatically computed variable calculation mechanisms to ensure transcription accuracy, consistency, and completion, as well as to ensure intra- and inter-observer reliability.

Data on demographics, injury details, and surgical management were recorded and included: age; sex; pre-fracture residence (coded as: Home/Sheltered Housing; Care/Nursing Home, or ‘Hospital’); injury date; location where injury was sustained (coded as: Home/Indoor; Outdoor, or Hospital); admission date; date of surgery; surgical procedure; surgical delay status (defined as being surgery out with 36 h of admission), and reason for nonoperative management (if applicable).

Data concerning clinical patient factors were recorded and included: American Society of Anesthesiologists (ASA) classification, presence of major comorbidity (cardiovascular disease, renal disease, pulmonary disease, dementia, active cancer, or diabetes mellitus) and laboratory blood tests taken on admission (haemoglobin concentration, lymphocyte count, platelet count, serum sodium concentration, and serum albumin concentration). These laboratory blood tests were included on the basis of existing evidence that they may correlate with either disease severity in COVID-19 specifically, or with outcomes in hip fracture patients.[13], [14], [15], [16], [17] The Nottingham Hip Fracture Score (NHFS) was calculated from the variables included in the dataset.

COVID-19 diagnosis

Data in relation to COVID-19 status in the preoperative and postoperative periods were collected independently and included whether patients demonstrated clinical features of COVID-19 infection, as well as any SARS-CoV-2 rt-PCR test result (positive or negative) obtained via the standard oropharyngeal and nasopharyngeal swab technique as part of the routine clinical management.

Outcomes

Data relevant to early patient outcome measures were collected and included: date and destination of discharge from acute admission (defined as the acute orthopaedic trauma admission, or the total acute hospital admission if a patient was transferred from an acute centre to another acute centre of comparable care level), date of death, and whether death occurred during the acute admission. Patients were followed up for a minimum of 30 days following presentation with hip fracture.

Statistical methods

Statistical analyses were performed using Statistical Product and Service Solutions version 17.0 (SPSS Inc. Released 2008. SPSS Statistics for Windows, Version 17.0. Chicago: SPSS Inc.). Parametric and non-parametric tests were used as appropriate to analyse continuous variables for significant differences between groups. Unpaired t-tests were used to compare values between groups for numerical variables that demonstrated a normal distribution. A Chi square test was used to assess dichotomous variables for differences between groups (Fisher's exact test was used if the frequency was 5 or less in any one cell). Kaplan–Meier methodology was used to investigate 30-day survival after hip fracture and Log rank was used to compare survival between patients who had a positive COVID-19 diagnosis with those with a negative COVID-19 diagnosis. Cox regression analysis was used to assess the independent association of COVID-19 status on 30-day mortality and factors associated with 30 day mortality in patients with COVID-19. Logistic regression analysis was used to assess the independence of predictors associated with a positive COVID-19 diagnosis. Receiver operating characteristic (ROC) curve analysis was used to identify a threshold values in the scalar variables that were identified as predictors associated with a positive COVID-19 diagnosis: i) on admission; ii) after admission, and iii) at any time. The area under the ROC curve (AUC) ranges from 0.5 (which indicates a test with no accuracy in distinguishing whether a patient is COVID-19-positive), to 1.0 (where the test accurately identifies all COVID-19-positive patients). The threshold value was defined as the point at which the sensitivity and specificity were maximal in predicting a COVID-19-positive patient. A p-value of <0.05 was defined as statistically significant.

Results

During the audit period data for 7387 patients with a hip fracture from 14 different countries were submitted. Data were excluded for 104 patients (1.4%) who were younger than 50 years of age or who presented outside the audit period. Another 193 patients (2.6%) did not have a COVID-19 status recorded and were excluded from further analysis (Fig. 1). The final cohort consisted of 7090 patients of whom 4959 (69.9%) were female and 2130 (30.0%) male (one patient did not have sex recorded). Mean age was 82.2 years (standard deviation (SD) 10.6, range 50–104) (Table 1 ).

Fig. 1

Flow chart showing all patients, included and excluded patients, mortality outcomes according to COVID-19 status, and distribution of patients from participating nations.

Table 1

Patient demographics, Nottingham hip fracture score, residence, place of injury, comorbidity, surgery within 36 h, ASA grade, surgical management, admission blood test and COVID status according to 30-day mortality.

Demographic	Descriptive	30-day Mortality		Difference/Odds Ratio (95% CI)	p-valuea
		Alive (n = 6438)	Dead (n = 652)
Age (years: mean, SD)		81.8 (10.7)	86.0 (9.0)	Diff 4.2 (3.3–5.0)	<0.001
Sex (n, % of group)	Female	4602 (71.48)	357 (54.75)	Reference
	Male	1836 (28.52)	294 (45.10)	2.06 (1.75–2.43)	<0.001
	Missing	0	1 (0.15)	N/A	–
Nottingham Hip Score (mean, SD)		4.8 (2.4)	6.0 (3.9)	Diff 1.2 (1.0–1.4)	<0.001
Residence (n, % of group)	Home/Sheltered	4975 (77.27)	390 (59.82)	Reference
	Care/Nursing home	1166 (18.11)	221 (56.67)	2.42 (1.03–2.89)	<0.001
	Hospital	81 (1.26)	22 (3.37)	3.46 (2.14–5.61)	<0.001
	Missing	216 (3.34)	19 (2.91)	1.12 (0.69–1.81)	0.639
Place of injury (n, % of group)	Home/Indoor	5082 (78.94)	552 (84.66)	Reference
	Outdoor	919 (14.27)	40 (6.13)	0.40 (0.29–0.56)	<0.001
	Hospital	154 (2.39)	37 (5.67)	2.21 (1.53–3.20)	<0.001
	Missing	283 (4.40)	23 (3.53)	0.75 (0.48–1.15)	0.188
Comorbiditya (n, % of group)	Not present	Reference
	CVD	4115 (63.92)	486 (74.54)	1.67 (1.39–2.01)	<0.001
	Renal Disease	1281 (19.90)	209 (3.25)	1.91 (1.60–2.27)	<0.001
	Pulmonary Disease	1362 (21.16)	216 (3.36)	1.85 (1.56–2.20)	<0.001
	Dementia	1868 (29.02)	284 (4.41)	1.90 (1.61–2.24)	<0.001
	Cancer	630 (9.79)	109 (1.69)	1.86 (1.49–2.32)	<0.001
	Diabetes Mellitus	1289 (20.02)	126 (1.96)	0.96 (0.78–1.18)	0.696
Surgery <36 h (n, % of group)	Yes	4043 (62.80)	338 (5.25)	Reference
	No	2253 (35.00)	214 (3.32)	1.14 (0.95–1.36)	0.162
	N/A	110 (1.71)	94 (1.46)	10.22 (7.60–13.75)	<0.001
	Missing	32 (0.50)	6 (0.09)	2.24 (0.93–5.40)	0.381
ASA grade (n, % of group)	1	118 (0.02)	4 (0.06)	1.48 (0.52–4.26)
	2	1400 (21.75)	32 (0.50)	Reference
	3	3720 (57.78)	354 (5.50)	4.15 (2.88–5.99)	<0.001
	4	945 (14.68)	219 (33.59)	10.14 (6.93–14.8)	<0.001
	5	5 (0.08)	16 (2.45)	13.67 (4.72–39.60)	<0.001
	Missing or N/A	250 (3.88)	27 (4.14)	4.73 (2.78–8.02)	<0.001
Management (n, % of group)	Fixation	3199 (49.69)	292 (44.78)	Reference
	Arthroplasty	3049 (47.36)	255 (39.11)	0.92 (0.77–1.09)	0.327
	Non-operative	104 (1.62)	91 (13.96)	9.59 (7.06–13.01)	<0.001
	Other	35 (0.54)	8 (1.23)	2.50 (1.15–5.45)
	Missing	51 (0.79)	6 (0.92)	1.29 (0.55–3.03)
Admission Blood Tests (mean, SD)
Haemoglobin Concentration (g/L)	n = 6435 vs 650	122.9 (18.0)	118.9 (19.8)	3.9 (2.5–5.4)	<0.001
Lymphocyte Count (x 109/L)	n = 6430 vs 650	1.21 (0.73)	1.09 (0.62)	0.12 (0.06–0.18)	<0.001
Platelet Count (x 109/L)	n = 6430 vs 648	245.8 (89.1)	243.8 (98.6)	2.0 (−5.2 to 9.3)	0.582
Sodium Concentration (mmol/L)	n = 6414 vs 648	137.6 (1.4)	137.6 (4.8)	0.0 (−0.3 to 0.4)	0.879
Albumin Concentration (g/L)	n = 6256 vs 641	36.6 (5.9)	33.8 (6.2)	2.8 (0.3–1.7)	0.006
COVID-19 status (n, % of group)	No	5965 (92.65)	474 (72.70)	Reference
	Yes	473 (7.35)	178 (27.30)	4.74 (3.89–5.76)	<0.001
	No	5965 (92.65)	474 (72.70)	Reference
	On admission	169 (2.62)	56 (8.59)	4.17 (3.04–5.72)	<0.001
	Postoperative	304 (4.72)	122 (18.71)	5.05 (4.01–6.36)	<0.001

Data not available for four patients: two died within the 30 day follow up period.

The independent influence of COVID-19 on patient mortality

There were 651 (9.2%) patients who were assigned a diagnosis of COVID-19, of whom 225 (34.6%) were positive preoperatively and 426 (65.4%) positive postoperatively. In total 652 (9.2%) patients died within and including 30 days of presentation with a hip fracture, of whom 178/652 (27.3%) had been diagnosed with COVID-19. Patients diagnosed with COVID-19 at any timepoint had a significantly lower 30-day survival rate when compared to those without COVID-19 (72.7%, 95% Confidence Interval (CI) 69.4 to 76.0% versus 92.6%, 95% CI 92.4 to 92.8, Log rank p < 0.001, Fig. 2 ). There was no significant difference in 30-day survival (Log rank p = 0.661) when comparing those diagnosed with COVID-19 preoperatively (75.1%, 95% CI 69.4 to 80.8) and those diagnosed postoperatively (71.4%, 95% CI 67.1 to 75.7); survival was significantly lower for both groups (Log rank p < 0.001) than for patients without COVID-19 (Fig. 3 ).

Fig. 2

Kaplan Meier curve for 30-day survival according to whether a patient was COVID negative (black) or COVID positive (red) within 30-days of admission. Log rank p < 0.001, 92.6% (95% CI 92.4 to 92.8) versus 72.7% (95% CI 69.4 to 76.0) at 30-days.

Fig. 3

Kaplan Meier curve for 30-day survival according to whether a patient was COVID negative (black), COVID positive at admission (red) or COVID positive after admission (grey). Log rank p = 0.661, between COVID positive patients preoperatively (75.1%, 95% CI 69.4 to 80.8) versus postoperatively (71.4%, 95% CI 67.1 to 75.7) at 30-days.

Unadjusted analysis of factors associated with increased 30-day mortality were older age (p < 0.001), male sex (p < 0.001), a higher Nottingham Hip Fracture Score (p < 0.001), care/nursing home (p < 0.001) or hospital (p < 0.001) residence, hip fracture sustained indoors or in hospital (p < 0.001), cardiovascular disease (<0.001), renal disease (p < 0.001), pulmonary disease (p = 0.012), dementia (p = 0.004), active cancer (p = 0.039), higher ASA grades (4 or 5) (p < 0.001), and a positive COVID-19 status (p < 0.001) (Table 1 ). The significant influence of non-operative management (p < 0.001) and consequent ‘not applicable’ classification regarding surgery within 36 h of admission (p < 0.001) on mortality (Table 1) was thought to be a secondary marker of increased mortality risk due to frailty and was thus not included in the regression models. Cox regression analysis (Table 2 ) identified that a diagnosis of COVID-19 was associated with a significantly increased mortality rate in the 30-days following admission for a hip fracture after adjusting for confounding factors (Hazard ratio (HR) 2.83, 95% CI 2.33 to 3.42, p < 0.001). The associated HR was higher if COVID-19 was diagnosed after admission (3.09, 95% CI 2.48 to 3.85) compared to those diagnosed on admission (2.36, 95% 1.73 to 3.21), but this was not statistically different.

Table 2

Cox regression model identifying patient related factors associated with 30-day mortality following a hip fracture.

Demographic	Descriptive	Hazard Ratio (95% CI)	p-value∗
Age (for each increasing year)		1.04 (1.03–1.05)	<0.001
Sex	Female	Reference
	Male	1.93 (1.63–2.30)	<0.001
Nottingham Hip Score (for each increasing point)		0.99 (0.96–1.01)	0.331
Residence	Home/Sheltered	Reference
	Care/Nursing home	1.44 (1.17–1.77)	0.001
	Hospital	1.23 (0.67–2.26)	0.507
	Missing	0.85 (0.52–1.40)	0.854
Place of injury	Home/Indoor	Reference
	Outdoor	0.65 (0.45–0.94)	0.022
	Hospital	1.20 (0.75–1.91)	0.452
	Missing	0.69 (0.40–1.18)	0.174
Comorbidity∗	Not present
	CVD	1.17 (0.96–1.42)	0.129
	Renal Disease	1.23 (1.02–1.48)	0.028
	Pulmonary Disease	1.45 (1.21–1.73)	<0.001
	Dementia	1.11 (0.91–1.35)	0.299
	Cancer	1.46 (1.16–1.85)	0.001
ASA grade	1	3.06 (1.06–8.78)	0.038
	2	Reference
	3	2.31 (1.55–3.45)	<0.001
	4	3.50 (2.30–5.32)	<0.001
	5	7.43 (3.65–15.12)	<0.001
	Missing or N/A	2.76 (1.58–4.81)	<0.001
Admission Blood Tests (for each increasing point)	Haemoglobin Concentration (g/L)	1.00 (0.99–1.01)	0.443
	Lymphocyte Count (x 109/L)	0.94 (0.83–1.07)	0.321
	Albumin Concentration (g/L)	0.96 (0.94–0.97)	<0.001
COVID-19 status	No	Reference
	Yes	2.83 (2.33–3.42)	<0.001
	Substituted in the model
	No	Reference
	On admission	2.36 (1.73–3.21)	<0.001
	Postoperative	3.09 (2.48–3.85)	<0.001

Predictors associated with having COVID-19 at any time

Factors associated with a positive COVID-19 status on unadjusted analysis were older age (p < 0.001), male sex (p = 0.012), a higher Nottingham Hip Fracture score (p = 0.001), place of residence (p = 0.001), place of injury (p = 0.001), cardiovascular disease (p = 0.001), renal disease (p = 0.039), pulmonary disease (p = 0.013), dementia (p = 0.001), active cancer (p = 0.046), increasing ASA grade (p < 0.001), lower lymphocyte count (p < 0.001), lower serum albumin concentration (p < 0.001) increased length of hospital stay (p < 0.001) (Table 3 ). Regression analysis demonstrated male sex, residence in a care/nursing home, place of injury, active cancer, ASA grade 4 and 5, and increased length of stay were independently associated with positive COVID-19 status (Table 4 ).

Table 3

Patient demographics, Nottingham hip fracture score, admission blood results, residence, place of injury, comorbidity, time to surgery, ASA grade, management, admission blood tests, length of stay, and mortality according to COVID status.

Demographic	Descriptive	COVID-19 Status		Difference/Odds Ratio (95% CI)	p-valuea
		Negative (n = 6439)	Positive (n = 651)
Age (years: mean, SD)		82.0 (10.7)	84.3 (9.0)	2.3 (1.5–3.2)	<0.001
Sex (n, % of group)	Female	4550 (70.66)	409 (0.15)	Reference
	Male	1888 (29.32)	242 (37.17)	1.43 (1.21–1.69)	<0.001
	Missing	1 (0.01)	0 (0.00)	N/A
Nottingham HipFractureScore (mean, SD)		4.8 (2.4)	5.6 (4.0)	0.8 (0.6–1.0)	<0.001
Residence (n, % of group)	Home/Sheltered	5004 (77.71)	361 (55.45)	Reference
	Care/Nursing home	1160 (18.01)	227 (34.87)	2.71 (2.27–3.24)	<0.001
	Hospital	83 (1.29)	20 (3.07)	3.34 (2.03–5.51)	<0.001
	Missing	192 (2.98)	43 (6.60)	3.10 (2.19–4.39)	<0.001
Place of injury (n, % of group)	Home/Indoor	5090 (79.05)	544 (83.56)	Reference
	Outdoor	916 (14.22)	43 (6.60)	0.44 (0.32–0.60)	<0.001
	Hospital	152 (2.36)	39 (5.99)	2.40 (1.67–3.45)	<0.001
	Missing	281 (4.36)	25 3.84)	0.83 (0.55–1.27)	0.390
Comorbiditya (n, % of group)	Not present
	CVD	4130 (64.14)	471 (72.35)	1.47 (1.23–1.76)	<0.001
	Renal Disease	1333 (20.70)	157 (24.12)	1.22 (1.01–1.48)	0.039
	Pulmonary Disease	1408 (21.87)	170 (26.11)	1.26 (1.05–1.52)	0.013
	Dementia	1865 (28.96)	287 (44.09)	1.94 (1.64–2.28)	<0.001
	Cancer	686 (10.65)	53 (8.14)	0.74 (0.56–1.0)	0.046
	Diabetes Mellitus	1277 (19.83)	138 (21.20)	1.09 (0.89–1.33)	0.398
Surgery <36 h (n, % of group)	Yes	3991 (61.98)	390 (59.91)	Reference
	No	2246 (34.88)	221 (33.95)	1.01 (0.85–1.20)	0.920
	N/A	167 (2.59)	37 (5.68)	2.27 (1.56–3.29)	<0.001
	Missing	35 (0.54)	3 (0.46)	0.88 (0.27–2.87)
ASA grade (n, % of group)	1	119 (1.85)	3 (0.46)	0.50 (0.15–1.61)	0.233
	2	1363 (21.17)	69 (10.60)	Reference
	3	3705 (57.55)	369 (56.68)	1.97 (1.51–2.56)	<0.001
	4	983 (15.27)	181 (27.80)	3.64 (2.72–4.85)	<0.001
	5	12 (0.19)	9 (1.38)	14.82 (6.04–36.35)	<0.001
	Missing or N/A	257 (3.99)	20 (3.07)	1.54 (0.92–2.57)	0.100
Management (n, % of group)	Fixation	3181 (49.40)	310 (47.62)	Reference
	Arthroplasty	3010 (46.75)	294 (45.16)	1.00 (0.86–1.16)	0.999
	Non-operative	160 (2.48)	35 (5.38)	2.24 (1.52–3.29)	<0.001
	Other	37 (0.57)	6 (0.92)	1.66 (0.69–3.97)
	Missing	51 (0.79s)	6 (0.92)	1.20 (0.51–2.83)	0.671
Admission Blood Tests (mean, SD)
Haemoglobin Concentration (g/L)	n = 6434 vs 651	122.6 (18.3)	121.5 (17.7)	1.1 (−0.3 to 2.6)	0.132
Lymphocyte Count (x 109/L)	n = 6425 vs 651	1.21 (0.72)	1.07 (0.68)	0.14 (0.08–0.19)	<0.001
Platelet Count (x 109/L)	n = 6427 vs 651	246.0 (90.0)	241.8 (89.8)	4.3 (−3.0 to 11.5)	0.250
Sodium Concentration (mmol/L)	n = 6411 vs 651	137.6 (4.4)	137.6 (4.7)	0.0 (−0.4 to 0.4)	0.919
Albumin Concentration (g/L)	n = 5546 vs 576	36.4 (6.0)	35.3 (5.8)	1.2 (0.7–1.7)	<0.001
LOS (days: mean, SD)		10.4 (7.7)	17.2 (13.1)	6.7 (6.0–7.4)	<0.001
30-day mortality (n, % of group)	No	5965 (92.64)	473 (72.66)	Reference
	Yes	474 (7.36)	178 (27.34)	4.74 (3.89–5.76)	<0.001

∗∗chi square test.

Unpaired Students t-test unless otherwise stated.

Table 4

Logistic regression model identifying patient related factors associated with COVID-19 positive patients and a hip fracture.

Demographic	Descriptive	Odds Ratio (95% CI)	p-value∗
Age (for each increasing year)		1.00 (0.99–1.02)	0.428
Sex	Female	Reference
	Male	1.38 (1.13–1.69)	0.001
Nottingham Hip Score (for each increasing point)		1.03 (0.99–1.06)	0.129
Residence	Home/Sheltered	Reference
	Care/Nursing home	2.15 (1.69–2.73)	<0.001
	Hospital	1.31 (0.63–2.72)	0.467
	Missing	2.57 (1.73–3.83)	<0.001
Place of injury	Home/Indoor	Reference
	Outdoor	0.58 (0.40–0.84)	0.004
	Hospital	2.23 (1.31–3.79)	0.003
	Missing	1.22 (0.74–2.01)	0.436
Comorbidity∗	Not present
	CVD	1.24 (0.99–1.53)	0.051
	Renal Disease	0.85 (0.68–1.07)	0.165
	Pulmonary Disease	0.99 (0.79–1.23)	0.917
	Dementia	1.18 (0.94–1.48)	0.164
	Cancer	0.63 (0.44–0.89)	0.009
ASA grade	1	0.69 (0.21–2.31)	0.548
	2	Reference
	3	1.16 (0.85–1.57)	0.352
	4	1.59 (1.13–2.25)	0.008
	5	8.28 (2.81–24.42)	<0.001
	Missing or N/A	0.68 (0.36–1.30)	0.246
Admission Blood tests (for each point)	Lymphocyte Count (x 109/L)	0.83 (0.71–0.98)	0.023
	Albumin Concentration (g/L)	0.99 (0.97–1.00)	0.102
Length of stay (for each increasing day)		1.06 (1.05–1.07)	<0.001

Predictors associated with having COVID-19 on admission

There were 225 patients who had COVID-19 at the time of presentation with hip fracture. Regression analysis demonstrated residence in a care/nursing home, in hospital fracture, ASA grade 5, lower lymphocyte count and albumin were all independently associated with a positive COVID-19 diagnosis on admission (Table 5 ). ROC curve analysis illustrated that a lymphocyte count at time of presentation of ≤0.93 and an albumin level of ≤36 g/dL were predictors of COVID-19 on admission (Fig. 4 ), but were poorly predictive, with an AUC of approximately 60%.

Table 5

Logistic regression model identifying patient related factors associated with COVID-19 positive patients on admission with a hip fracture.

Demographic	Descriptive	Odds Ratio (95% CI)	p-value∗
Age (for each increasing year)		1.00 (0.99–1.02	0.843
Sex	Female	Reference
	Male	1.01 (0.71–1.50)	0.941
Nottingham Hip Score (for each increasing point)		0.98 (0.82–1.19)	0.862
Residence	Home/Sheltered	Reference
	Care/Nursing home	4.13 (2.78–6.13)	<0.001
	Hospital	0.85 (0.31–2.35)	0.851
	Missing	0.54 (0.13–1.26)	0.400
Place of injury	Home/Indoor	Reference
	Outdoor	0.52 (0.25–1.09)	0.085
	Hospital	4.98 (2.64–9.38)	<0.001
	Missing	0.71 (0.22–2.28)	0.561
Comorbidity∗	Not present	Reference
	CVD	0.96 (0.69–1.33)	0.800
	Renal Disease	0.78 (0.54–1.14)	0.202
	Pulmonary Disease	0.87 (0.61–1.26)	0.471
	Dementia	1.24 (0.81–1.92)	0.324
	Cancer	0.61 (0.33–1.13)	0.117
ASA grade	1	1.43 (0.32–6.34)	0.636
	2	Reference
	3	0.97 (0.60–1.57)	0.902
	4	1.47 (0.86–2.51)	0.159
	5	5.25 (1.30–21.31)	0.020
	Missing or N/A	0.58 (0.23–1.49)	0.258
Admission Blood Tests (for each point)	Lymphocyte Count (x 109/L)	0.62 (0.46–0.83)	0.001
	Albumin (g/L)	0.95 (0.93 0.98)	<0.001

Fig. 4

ROC curve for lymphocyte count (grey) and albumin (black dashed) as a predictor of COVID-19 on admission. Lymphocyte: Area under the curve 60.7% (95% CI 56.7%–64.6%, p < 0.001). Threshold of 0.93 or less has 58.2% specificity and 56.6% sensitivity. Albumin: Area under the curve 61.3% (95% CI 57.5%–65.2%, p < 0.001). Threshold of 36 g/dL or less has 59.1% specificity and 57.1%sensitivity.

Predictors associated with having COVID-19 after admission

There were 426 patients diagnosed with positive COVID-19 after admission to hospital. Regression analysis demonstrated male sex, a fall indoor, cardiovascular disease, ASA grade 4 or 5, and longer duration of hospital stay were independently associated with a positive COVID-19 diagnosis on admission (Table 6). ROC curve analysis illustrated that length of stay of 10 or more days was a moderately reliable predictor of COVID-19 following admission (Fig. 5 ), with an AUC of 71.6%.

Table 6

Logistic regression model identifying patient related factors associated with developing COVID-19 in hip fracture patients following admission.

Demographic	Descriptive	Odds Ratio (95% CI)	p-value∗
Age (for each increasing year)		1.01 (0.99–1.02)	0.480
Sex	Female	Reference
	Male	1.56 (1.23–1.97)	<0.001
Nottingham Hip Score (for each increasing point)		1.03 (0.99–1.06)	0.110
Residence	Home/Sheltered	Reference
	Care/Nursing home	1.22 (0.89–1.67)	0.218
	Hospital	2.03 (0.81–5.11)	0.133
	Missing	3.14 (2.07–4.77)	<0.001
Place of injury	Home/Indoor	Reference
	Outdoor	0.56 (0.36–0.87)	0.009
	Hospital	1.03 (0.79–2.36)	0.942
	Missing	1.37 (0.79–2.36)	0.263
Comorbidity∗	Not present
	CVD	1.43 (1.09–1.86)	0.009
	Renal Disease	0.90 (0.69–1.18)	0.433
	Pulmonary	1.03 (0.79–1.34)	0.850
	Dementia	1.18 (0.89–1.55)	0.254
	Cancer	0.65 (0.43–0.98)	0.041
ASA grade	1	0.36 (0.05–2.69)	0.317
	2	Reference
	3	1.35 (0.92–1.97)	0.123
	4	1.79 (1.16–2.75)	0.008
	5	10.84 (3.09–38.00)	<0.001
	Missing or N/A	0.69 (0.29–1.62)	0.394
Admission Blood Tests (for each point)	Lymphocyte Count (x 109/L)	0.92 (0.77–1.10)	0.383
	Albumin (g/L)	1.00 (0.99–1.08)	0.681
Length of stay (for each increasing day)		1.07 (1.06–1.08)	<0.001

Fig. 5

ROC curve for length of hospital stay (dashed line) as a predictor of developing COVID-19 following admission. Area under the curve 71.6% (95% CI 68.8%–74.4%, p < 0.001). Threshold of 10 days or more has 65% specificity and sensitivity.

Predictors associated with increased mortality in patients with COVID-19

Factors associated with increased risk of 30-day mortality on unadjusted analysis were older age, male sex, higher NHFS, injury sustained outdoors, renal disease, pulmonary disease, dementia, increasing ASA grade, nonoperative management, lower lymphocyte count, lower platelet count, and lower serum albumin concentration (Table 7 ). Regression analysis demonstrated that increasing age (HR 1.03, 95% CI 1.01–1.05, p = 0.028), male sex (HR 2.35, 95% CI 1.66–3.34, p < 0.001), renal disease (HR 1.53, 95% CI 1.08–2.18, p = 0.017), and pulmonary disease (HR 1.45, 95% CI 1.02–2.06, p = 0.039) were independently associated with an increased risk of 30-day mortality (Table 8 ).

Table 7

Patient demographics, Nottingham hip fracture score, residence, place of injury, comorbidity, surgery within 36 h, ASA grade, surgical management, admission blood test according to 30-day mortality for COVID-19 positive patients only.

Demographic	Descriptive	30-day Mortality		Difference/Odds Ratio (95% CI)	p-valuea
		Alive (n = 473)	Dead (n = 178)
Age (years: mean, SD)		83.7 (9.5)	85.8 (7.5)	Diff 2.1 (0.5–3.7)	0.008
Sex (n, % of group)	Female	326	83	Reference
	Male	147	95	OR 2.54 (1.78–3.61)	<0.001
	Missing	0	0
Nottingham Hip Score (mean, SD)		5.3 (1.6)	6.5 (7.1)	Diff 1.2 (0.6–1.9)	<0.001
Residence (n, % of group)	Home/Sheltered	270	91	Reference
	Care/Nursing home	154	73	OR 1.41 (0.98–2.03)	0.067
	Hospital	15	5	OR 0.99 (0.45–2.16)	0.999
	Missing	34	9	OR 0.83 (0.45–1.52)	0.537
Place of injury (n, % of group)	Home/Indoor	385	159	Reference
	Outdoor	38	5	OR 0.32 (0.12–0.82)	0.013
	Hospital	30	9	OR 0.73 (0.34–1.56)	0.413
	Missing	20	5	OR 0.61 (0.22–1.64)	0.375
Comorbiditya (n, % of group)	Not present	Reference		Reference
	CVD Disease	335	136	OR 1.33 (0.90–1.99)	0.156
	Renal Disease	96	61	OR 2.04 (1.39–2.99)	<0.001
	Pulmonary Disease	109	61	OR 1.74 (1.20–2.54)	0.004
	Dementia	196	91	OR 1.48 (1.05–2.09)	0.027
	Cancer	37	16	OR 1.16 (0.63–2.15)	0.628
	Diabetes Mellitus	104	34	OR 0.83 (0.54–1.29)	0.645
Surgery <36 h (n, % of group)	Yes	288	102	Reference
	No	173	48	OR 0.78 (0.53–1.16)	0.221
	N/A	10	27	OR 7.62 (3.57–16.30)	<0.001
	Missing	2	1	OR 1.41 (0.13–15.74)	0.999
ASA grade (n, % of group)	1	2	1	OR 6.40 (0.49–83.39)	0.233
	2	64	5	Reference
	3	271	98	OR 4.63 (1.81–11.84)	<0.001
	4	120	61	OR 6.51 (2.49–17.01)	<0.001
	5	1	8	OR 102.40 (10.59–990.6)	<0.001
	Missing or N/A	15	2	OR (1.71 90.30 to 9.66)	0.621
Management (n, % of group)	Fixation	225	85	Reference
	Arthroplasty	227	67	0.78 (0.54–1.13)	0.190
	Non-operative	10	25	6.62 (3.05–14.36)	<0.001
	Other	6	0	–	0.197
	Missing	5	1	0.53 (0.06–4.60)	0.685
Admission Blood Tests (mean, SD)
Haemoglobin	n = 473 vs 178	121.7 (17.4)	120.8 (18.4)	0.9 (−2.1 to 4.0)	0.558
Lymphocyte	n = 473 vs 178	1.11 (0.67)	0.98 (0.70)	0.13 (0.01–0.25)	0.030
Platelet	n = 473 vs 178	245.7 (91.5)	231.3 (84.7)	14.5 (−1.0 to 29.9)	0.067
Sodium	n = 473 vs 178	137.5 (4.7)	138.0 (4.7)	0.6 (−0.3 to 1.4)	0.180
Albumin	n = 419 vs 157	34.4 (5.7)	35.6 (5.8)	1.2 (0.1–2.3)	0.027
Time of COVID-19 Diagnosis (n, % of group)	Admission	169	56	Reference
	Following admission	304	122	1.21 (0.84–1.75)	0.307

Data not available for four patients: two died within the 30 day follow up period.

Table 8

Cox regression model identifying patient related factors associated with 30-day mortality following a hip fracture in patients for patients with COVID-19.

Demographic	Descriptive	Hazard Ratio (95% CI)	p-value∗
Age (for each increasing year)		1.03 (1.01–1.05)	0.028
Sex	Female	Reference
	Male	2.35 (1.66–3.34)	<0.001
Nottingham Hip Score (for each increasing point)		1.00 (0.97–1.03	0.825
Residence	Home/Sheltered	Reference
	Care/Nursing home	1.32 (0.90–1.95)	0.155
	Hospital	1.17 (0.30–4.45)	0.823
	Missing	0.98 (0.46–2.12)	0.982
Place of injury	Home/Indoor	Reference
	Outdoor	0.35 (0.11–1.14)	0.081
	Hospital	0.64 (0.24–1.72)	0.374
	Missing	0.32 (0.06–1.56)	0.158
Comorbidity	Not present	Reference
	Renal Disease	1.53 (1.08–2.18)	0.017
	Pulmonary	1.45 (1.02–2.06)	0.039
	Dementia	1.24 (0.85–1.83)	0.266
ASA grade	1	8.69 (0.96–78.75)	0.055
	2	Reference
	3	2.36 (0.94–5.88)	0.066
	4	2.41 (0.94–6.14)	0.066
	5	2.66 (0.78–9.02)	0.117
	Missing or N/A	1.97 (0.46–8.44)	0.358
Management	Fixation	Reference
	Arthroplasty	0.75 (0.53–1.06)	0.103
	Non-operative	2.59 (1.52–4.43)	<0.001
	Other	–
	Missing	1.29 (0.13–12.38)	0.824
Blood tests (for each increasing unit)	Lymphocyte	0.83 (0.62–1.12)	0.233
	Platelet	1.00 (1.00–1.00)	0.085
	Albumin	0.98 (0.95–1.01)	0.132

Discussion

This global multicentre audit reports the findings from 112 hospitals in 14 countries. A positive diagnosis of COVID-19 during an acute admission for hip fracture was independently associated with an approximate three-fold increase in 30-day mortality risk compared to patients without COVID-19, and it is likely that hip fracture patients are the single group of surgical admissions that account for the largest number of COVID-19-related deaths. Approximately two thirds of COVID-19 cases were diagnosed postoperatively, which supports findings from a previous study suggesting the major role of nosocomial transmission among this vulnerable patient group. For the first time, clinical factors that are associated with increased risk of death in hip fracture patients who have COVID-19 are reported and this may help to identify fragility trauma patients that could benefit from isolating or shielding. This study, which is understood to be the largest multicentre orthopaedic collaborative audit delivered, offers the only global data into hip fracture and COVID-19 from the pre-vaccination era and could be used to ensure better preparedness for future disease outbreaks, from seasonal influenza to emerging diseases.

The prevalence of COVID-19 in this study cohort was 9.2%. This is consistent with the existing literature from single-centre or regional studies, but was many times higher than the mean background prevalence in any of the participating nations throughout the study period (range 0·0-0·5%). The extreme vulnerability of this patient group may be under-recognised among healthcare professionals, and the major disruption to fragility trauma services experienced globally is likely to contribute to an enduring public health crisis. Although the study investigated only patients with hip fracture, these findings are likely to be generalisable to frail trauma patients, as well as to the wider frail inpatient population.

The current data suggests that two-thirds of COVID-19 cases were diagnosed postoperatively, and IMPACT-Scot 2 demonstrated that approximately 60% of COVID-19 cases were likely to be hospital-acquired, with the majority of these nosocomial infections occurring in acute orthopaedic wards or following discharge to inpatient orthopaedic rehabilitation facilities. Nosocomial infection may be an important factor in the high rates of COVID-19 observed among vulnerable inpatients and this problem has significant implications for the spread of COVID-19 between hospitals, downstream bed facilities, residential care settings and the community. There remains little published evidence that demonstrates successful strategies for the mitigation of this phenomenon among frail orthogeriatric trauma patients.

The factors identified in the current study that were independently associated with a positive COVID-19 diagnosis (at any time) were consistent with the existing literature, although the current data identified differences depending on whether COVID-19 was identified at initial presentation or following admission, which is of particular relevance to clinical risk stratification and the isolation of at-risk patients. , Factors predictive of having COVID-19 at admission were certain admission laboratory blood tests (lower blood albumin level and lymphocyte count), higher pre-fracture care demands (residential or inpatient care) and a high ASA grade. Male sex, pre-existing cardiovascular disease, high ASA grade, and a longer length of stay were predictive of COVID-19 diagnoses made postoperatively. Most of these factors are indicators of increasing frailty and may indicate vulnerability to infection. These findings may assist stratification of patients according to their risk of transmitting or acquiring COVID-19 in hospital, and facilitate deployment of clinical patient pathways for isolating, shielding, or ‘cohorting’ patients in COVID and non-COVID circuits – an approach which has been found to be effective in the management of hip fracture patients during the pandemic. The key modifiable risk factor identified was length of stay, which supports previous work in this area that underlines that safeguarding and prioritisation of fragility fracture services as essential to help protect this vulnerable patient group through early treatment and discharge planning.[22], [23] However, the causal relationship of increased length of stay on the likelihood of contracting COVID-19 is difficult to determine, since patients with COVID-19 are likely to require a longer hospital admission, and frailer patients (who are more vulnerable to acquiring COVID-19) typically require longer inpatient management prior to discharge.

Male sex was associated with a two-fold increased risk of 30-day mortality among patients diagnosed with COVID-19. This supports existing evidence from the general population that males with COVID-19 have a higher mortality rate than females. Various explanatory mechanisms have been suggested and include differences in expression of angiotensin-converting enzyme II, smoking status, obesity, and behavioural factors.[25], [26], [27], [28] The existence of underlying pulmonary disease was independently associated with a higher 30-day mortality risk, which is consistent with the known pathophysiology of COVID-19. The influence of renal disease on mortality is of particular importance in hip fracture patients given the relatively high prevalence of chronic kidney disease, acute kidney injury, or mixed acute kidney injury and chronic kidney disease, all of which have been shown to be associated with poorer outcomes in non-hip fracture groups with COVID-19. The identification of these clinical predictors in the hip fracture population is original and could guide clinical decision-making and prognosis.

The COVID-19 pandemic remains a dynamic situation subject to: further increases in the incidence of SARS-CoV-2 infection; new viral strains with higher transmissibility, mortality risk, and resistance to vaccinations; the need to reduce restrictions in order to meet the needs of the population, and challenges associated with achieving widespread and effective vaccination across the globe. , [29], [30], [31] This study will provide an important baseline against which to measure factors such as vaccine efficacy, strategies for the mitigation of viral transmission, and the effects of different viral strains on this vulnerable population.

Evidence from the IMPACT collaborative has demonstrated widespread disruption to orthopaedic services, with resources and staff being repurposed for non-orthopaedic patients and standard operating procedures being overhauled in favour of other services. Hip fracture patients were managed on open generalist wards by non-specialised staff, experienced delays to surgery and appropriate care, received less specialist multidisciplinary management, and were exposed to an increase in inter-departmental transit. These issues are known to increased risk of nosocomial infection, delirium, and longer duration of hospital stay. , , In future communicable disease outbreaks it would be prudent to ensure the protection of specialist multidisciplinary teams, clinical areas, and access to prompt surgical management in line with existing standards of care for this most vulnerable patient group, as well as robust strategies to minimise in-hospital transmission through the use of clinical pathways and closed circuits that have previously been described. , , [33], [34], [35]

Early in the pandemic there was uncertainty about the infection prevention and control precautions required in the management of patients at risk of contracting SARS-CoV-2 infection. This caused disparities and frequent amendments to guidance about personal protective equipment, testing of patients and staff, the acceptability of risk relating to aerosol generating procedures such as cardiopulmonary resuscitation and anaesthetic procedures, and surgery. This led to confusion and delays to appropriate patient management and care ought to be taken to design procedures for the continuation of orthopaedic services in the context of future disease outbreaks. This is of relevance to unscheduled care and to urgent planned care, since the disruption has been to the detriment of patients attempting to access urgent elective care.[37], [38], [39]

The concerning finding of a high proportion of patients acquiring COVID-19 in the inpatient and downstream hospital settings raises questions regarding the efficacy of existing pathways and strategies for the prevention of infection transmission between healthcare services. The establishment of a robust and effective inpatient and post-discharge track and trace system could identify patients at risk of acquiring or transmitting infection, which has the potential to limit the harm from outbreaks and reduce the burden on rehabilitation and community health services.

This international study was conducted within the context of a rapidly-developing global pandemic. As a result, there are limitations inherent in the natural variation between nations relating to the background COVID-19 prevalence, which ranged from 0.003 to 0.294% during the study period. There was no standardised diagnostic protocol, such as routine regular testing of all patients, and the availability of laboratory testing may have varied between regions; the prevalence of COVID-19 may therefore have been underestimated. Furthermore, as routine clinical testing was not in place in most countries during the first wave of the pandemic, the mortality associated with undiagnosed COVID-19 was not quantifiable, and because the precise dates of COVID-19 diagnoses are not known the distinction between community- and hospital-acquired SARS-CoV-2 infections cannot be determined with certainty. This reflects real-world uncertainty around clinical criteria for diagnosing COVID-19 and variation in the approaches to population screening and symptomatic testing, and highlights the need to establish early consensus on these matters early in an outbreak in order to facilitate effective research and audit. There was variation in the approach to the provision of hip fracture services, though this could be considered a strength due to increased generalisability across the range of nations affected by the disease. Clinical audit in future outbreaks should strive for even greater coverage of geographical and health-economic context.[40], [41] Follow-up period was limited to 30 days post-presentation with hip fracture, which may underestimate mortality especially in patients who developed COVID-19 later in the admission. This limited follow-up is common amongst studies reporting the mortality associated with COVID-19. However, the current study controlled for this issue by reporting subgroups of patients with COVID-19 confirmed at initial presentation in the preoperative period versus later in the admission following surgical management. Variation in the systems available to clinicians to follow up patients after discharge may underestimate mortality rates in regions that don't have, for example, a unified healthcare system with patients linked by a universally-applied unique community identifier. This ought to be considered in the methodology of future studies. There remains a lack of evidence pertaining to the indirect effects of the pandemic on COVID-19-negative hip fracture, or the effect that mass population vaccination will have on prevalence, transmissibility, and mortality. There was heterogeneity in the literature reporting investigations in COVID-19 in hip fracture, with many studies being limited by a lack of robust diagnostic criteria, insufficient follow-up durations, unadjusted mortality analyses, and a lack of relevant information pertaining to background prevalence, pathogen variant profiles, and infection prevention and control measures in the catchment population. Adoption of shared reporting standards may improve the quality of evidence available to clinicians and researchers (Fig. 6 ).

Fig. 6

Suggested reporting standards for studies investigating COVID-19 in hip fracture patients.

The strengths of the study include the large number of patients and the unique international nature that has provided an analysis across a range of hospitals, hip fracture services, healthcare systems, ethnicities and reporting processes. This diversity would suggest that the findings are generalisable globally. The findings pertaining to COVID-19 prevalence, mortality risk, and predictors of infection support existing evidence and provide insight into clinical factors associated with COVID-19 and outcome. The high levels of participation in the UK and Spain in particular, ensured extensive coverage across these geographical areas, which may have helped account for regional variations in clinical practice, patient demographics and COVID-19 prevalence. Furthermore, the size of the COVID-19 positive cohort was large and afforded the first opportunity to perform subgroup regression analyses to identify factors associated with acquiring the infection and the mortality associated with it. The lessons learned from this study of the COVID-19 pandemic are applicable to future disease outbreaks and may facilitate better preparedness for other transmissible diseases such as seasonal influenza, emerging strains of existing pathogens, or novel communicable diseases.

Conclusion

The prevalence of COVID-19 in the hip fracture population was at least ten times higher than the background prevalence and was independently associated with a three-fold increase in 30-day mortality. Thus, hip fracture patients may be the cohort of hospital admissions that account for the largest number of COVID-19-related deaths. It is likely that nosocomial transmission of this disease was responsible for a significant proportion of infections, and the development of robust infection prevention and control strategies are likely to improve the management of future outbreaks. The IMPACT collaborative has demonstrated important lessons in the conduct of rapid clinical audit in order to guide the evidence-based response to emerging diseases, and a number of strategies are suggested that can be applied prospectively to ensure better preparedness for future health crises.

Funding

None.

Previous presentation of findings

This work was conducted in the context of an evolving global pandemic and the need for timely dissemination of information was critical. To this end a limited number of findings from the current study have been presented as abstracts at the British Orthopaedic Association Annual Congress 2021 (Free Paper Session: Infection & COVID-19), and the Scottish Committee for Orthopaedics and Trauma (SCOT) 2021 Meeting (Free Paper Session). ,

Declaration of competing interest

The authors declare that they have no conflict of interest.

IMPACT Global Group

Surname	Forename
Abdul-Jabar	Hani
Abu-Rajab	Rashid
Abugarja	Ahmed
Adam	Karen
Aguado Hernández	Héctor J.
Améstica Lazcano	Gedeón
Anderson	Sarah
Ansar	Mahmood
Antrobus	Jonathan
Aragón Achig	Esteban Javier
Archunan	Maheswaran
Arrieta Salinas	Mirentxu
Ashford-Wilson	Sarah
Assens Gibert	Cristina
Athanasopoulou	Katerina
Awadelkarim	Mohamed
Baird	Stuart
Bajada	Stefan
Balakrishnan	Shobana
Balasubramanian	Sathishkumar
Ballantyne	James A.
Bárcena Goitiandia	Leopoldo
Barkham	Benjamin
Barmpagianni	Christina
Barres-Carsi	Mariano
Barrett	Sarah
Baskaran	Dinnish
Bell	Jean
Bell	Katrina
Bell	Stuart
Bellelli	Giuseppe
Benchimol	Javier Alberto
Boietti	Bruno Rafael
Boswell	Sally
Braile	Adriano
Brennan	Caitlin
Brent	Louise
Brooke	Ben
Bruno	Gaetano
Burahee	Abdus
Burns	Shirley
Calabrò	Giampiero
Campbell	Lucy
Carabelli	Guido Sebastian
Carnegie	Carol
Carretero Cristobal	Guillermo
Caruana	Ethan
Cassinello Ogea	M.ª Concepción
Castellanos Robles	Juan
Castillon	Pablo
Chakrabarti	Anil
Cecere	Antonio Benedetto
Chen	Ping
Clarke	Jon V.
Collins	Grace
Corrales Cardenal	Jorge E.
Corsi	Maurizio
Cózar Adelantado	Gara María
Craxford	Simon
Crooks	Melissa
Cuarental-García	Javier
Cuthbert	Rory
Dall	Graham
Daskalakis	Ioannis
De Cicco	Annalisa
de la Fuente de Dios	Diana
Demaria	Pablo
Dereix	John
Díaz Jiménez	Julian
Dinamarca Montecinos	José Luis
Do Le	Ha Phuong
Donoso Coppa	Juan Pablo
Drosos	Georgios
Duffy	Andrew
East	Jamie
Eastwood	Deborah
Elbahari	Hassan
Elias de Molins Peña	Carmen
Elmamoun	Mamoun
Emmerson	Ben
Escobar Sánchez	Daniel
Faimali	Martina
Farré-Mercadé	Maria Victòria
Farrow	Luke
Fayez	Almari
Fell	Adam
Fenner	Christopher
Ferguson	David
Finlayson	Louise
Flores Gómez	Aldo
Freeman	Nicholas
French	Jonathan
Gabardo Calvo	Santiago
Gagliardo	Nicola
Garcia Albiñana	Joan
García Cruz	Guillermo
García de Cortázar Antolín	Unai
García Virto	Virginia
Gealy	Sophie
Gil Caballero	Sandra Marcela
Gill	Moneet
González González	María Soledad
Gopireddy	Rajesh
Guntley	Diane
Gurung	Binay
Guzmán Rosales	Guadalupe
Haddad	Nedaa
Hafeez	Mahum
Haller	Petra
Halligan	Emer
Hardie	John
Hawker	Imogen
Helal	Amr
Herrera Cruz	Mariana
Herreros Ruiz-Valdepeñas	Ruben
Horton	James
Howells	Sean
Howieson	Alan
Hughes	Luke
Hünicken Torrez	Flavia Lorena
Hurtado Ortega	Ana
Huxley	Peter
Hamid	Hytham K. S.
Ilahi	Nida
Iliadis	Alexis
Inman	Dominic
Jadhao	Piyush
Jandoo	Rajan
Jawad	Lucy
Jayatilaka	Malwattage Lara Tania
Jenkins	Paul J.
Jeyapalan	Rathan
Johnson	David
Johnston	Andrew
Joseph	Sarah
Kapoor	Siddhant
Karagiannidis	Georgios
Karanam	Krishna Saga
Kattakayam	Freddy
Konarski	Alastair
Kontakis	Georgios
Labrador Hernández	Gregorio
Lancaster	Victoria
Landi	Giovanni
Le	Brian
Liew	Ignatius
Logishetty	Kartik
Lopez Marquez	Andrew Carlomaria Daniel
Lopez	Judit
Lum	Joann
Macpherson	Gavin J.
Madan	Suvira
Mahroof	Sabreena
Malik-Tabassum	Khalid
Mallina	Ravi
Maqsood	Afnan
Marson	Ben
Martin Legorburo	M José
Martin-Perez	Encarna
Martínez Jiménez	Tania
Martinez Martin	Javier
Mayne	Alistair
Mayor	Amy
McAlinden	Gavan
McLean	Lucille
McDonald	Lorna
McIntyre	Joshua
McKay	Pamela
McKean	Greg
McShane	Heather
Medici	Antonio
Meeke	Chelsea
Meldrum	Evonne
Mendez	Mijail
Mercer	Scott
Merino Perez	Josu
Mesa-Lampré	María-Pilar
Mighton	Shuna
Milne	Kirsty
Mohamed Yaseen	Muhammed
Moppett	Iain
Mora	Jesus
Morales-Zumel	Sira
Moreno Fenoll	Irene Blanca
Mousa	Adham
Murray	Alastair W.
Murray	Elspeth V.
Nair	Radhika
Neary	Fiona
Negri	Giacomo
Negus	Oliver
Newham-Harvey	Fiona
Ng	Nigel
Nightingale	Jess
Noor Mohamed Anver	Sumiya
Nunag	Perrico
OHare	Matthew
Ollivere	Ben
Ortés Gómez	Raquel
Owens	AnneMarie
Page	Siobhan
Palloni	Valentina
Panagiotopoulos	Andreas
Panagiotopoulos	Elias
Panesar	Paul
Papadopoulos	Antonios
Spyridon	Papagiannis
Pareja Sierra	Teresa
Park	Chang
Parwaiz	Hammad
Paterson-Byrne	Paul
Patton	Sam
Pearce	Jack
Porter	Marina
Pellegrino	Achille
Pèrez Cuellar	Arturo
Pezzella	Raffaele
Phadnis	Ashish
Pinder	Charlotte
Piper	Danielle
Powell-Bowns	Matilda
Prieto Martín	Rocío
Probert	Annabel
Ramesh	Ashwanth
Ramírez de Arellano	Manuel Vicente Mejía
Renton	Duncan
Rickman	Stephen
Robertson	Alastair
Roche Albero	Adrian
Rodrigo Verguizas	José Alberto
Rodríguez Couso	Myriam
Rooney	Joanna
Sáez-López	Pilar
Saldaña-Díaz	Andres
Santulli	Adriano
Sanz Pérez	Marta Isabel
Sarraf	Khaled M.
Scarsbrook	Christine
Scott	Chloe E. H.
Scott	Jennifer
Shah	Sachi
Sharaf	Sharief
Sharma	Sidharth
Shirley	Denise
Siano	Antonio
Simpson	James
Singh	Abhinav
Singh	Amit
Sinnett	Tim
Sisodia	Gurudatt
Smith	Philomena
Sophena Bert	Eugenia
Steel	Michael
Stewart	Avril
Stewart	Claire
Sugand	Kapil
Sullivan	Niall
Sweeting	Lauren
Symes	Michael
Tan	Dylan Jun Hao
Tancredi	Francesco
Tatani	Irini
Thomas	Philip
Thomson	Fraser
Toner	Niamh S.
Tong	Anna
Toro	Antonio
Tosounidis	Theodoros
Tottas	Stylianos
Trinidad Leo	Andrea
Tucker	Damien
Vemulapalli	Krishna
Ventura Garces	Diego
Vernon	Olivia Katherine
Viveros Garcia	Juan Carlos
Ward	Alex
Ward	Kirsty
Watson	Kate
Weerasuriya	Thisara
Wickramanayake	Udara
Wilkinson	Hannah
Windley	Joseph
Wood	Janet
Wynell-Mayow	William
Zatti	Giovanni
Zeiton	Moez
Zurrón Lobato	Miriam