Hip fracture and pressure ulcers – the Pan‐European Pressure Ulcer Study – intrinsic and extrinsic risk factors

Abstract

Pressure ulcers (PU) in patients with hip fracture remain a problem. Incidence of between 8·8% and 55% have been reported. There are few studies focusing on the specific patient‐, surgery‐ and care‐related risk indicators in this group. The aims of the study were

• • to investigate prevalence and incidence of PU upon arrival and at discharge from hospital and to identify potential intrinsic and extrinsic risk factors for development of PU in patients admitted for hip fracture surgery.
• • to illuminate potential differences in patient logistics, surgery, PU prevalence and incidence and care between Northern and Southern Europe. Consecutive patients with hip fracture in six countries, Sweden, Finland, UK (North) and Spain, Italy and Portugal (South), were included. The patients were followed from Accident and Emergency Department and until discharge or 7 days. Prevalence, PU at discharge and incidence were investigated, and intrinsic and extrinsic risk indicators, including waiting time for surgery and duration of surgery were recorded. Of the 635 patients, 10% had PU upon arrival and 22% at discharge (26% North and 16% South). The majority of ulcers were grade 1 and none was grade 4. Cervical fractures were more common in the North and trochanteric in the South. Waiting time for surgery and duration of surgery were significantly longer in the South. Traction was more common in the South and perioperative warming in the North. Risk factors of statistical significance correlated to PU at discharge were age ≥71 (P = 0·020), dehydration (P = 0·005), moist skin (P = 0·004) and total Braden score (P = 0·050) as well as subscores for friction (P = 0·020), nutrition (P = 0·020) and sensory perception (P = 0·040). Comorbid conditions of statistical significance for development of PU were diabetes (P = 0·005) and pulmonary disease (P = 0·006). Waiting time for surgery, duration of surgery, warming or non warming perioperatively, type of anaesthesia, traction and type of fracture were not significantly correlated with development of PU.

Introduction

It is anticipated that the expected increase in the size of the elderly population of Europe will result in a significant increase in the numbers of patients hospitalised for hip fractures (1). In fact, during the past 20 years, the number of hip fractures in patients ≥80 years has doubled (2).

Pressure ulcers (PU) is a frequent complication to hip fractures, and incidence of between 8·8% and 55% has been reported 3, 4. Versluysen (1985) (5) reported that 17% of patients admitted for hip fracture surgery had PU upon arrival and that 34% developed lesions during the first week in hospital. The largest number of PU in this study occurred on the day of surgery. In another study by the same author (6), 66% of the patients with hip fracture developed PU, the majority of them appeared during the first 48 hours after admission. Pressure ulcers can develop at any time during a care episode 7, 8, but the majority of surgery‐related PU are reported to appear within 2–4 days post‐surgery 9, 10. Most PU are avoidable (11), but not all (12).

Pressure ulcers have a major impact on quality of life and on the cost of hospital care (13). Patients with PU report ‘endless pain’, restricted activities and impact on their family lives (14). Patients with p. u. require significantly more nursing time, remain hospitalised for longer periods of time, generate higher hospital charges and require more health care resources after discharge than comparable patients without PU 15, 16, 17, 18, 19. It has been reported that patients developing a PU spend an average of 33 days longer at a rehabilitation ward compared with patients without such ulcers (20). Another consequence of PU is death. In one study (21), a 35% 3‐month mortality for patients with PU was reported. Reduction in health‐related quality of life because of PU has also been reported 22, 23. There are a few studies attempting to identify specific risk factors for development of PU after hip fracture surgery. Some are retrospective 3, 24 and some prospective (4, 9, 25, 26). A retrospective review of medical records of hip fracture patients at 20 hospitals in the United states of America between 1983 and 1993 showed that wait for surgery, intensive unit stay, longer surgical procedure and general anaesthesia were significant risk factors for PU (3). However, since the nineties, awareness of PU risks and quality of preventive care has improved. General anaesthesia is now more uncommon in hip fracture surgery, and operation tables are usually well padded to avoid development of PU. Few patients are admitted to intensive care unit after hip fracture surgery.

There is an increasing consensus that PU are a sign of acute illness and that particular intrinsic and extrinsic factors affect the tolerance of skin to pressure. They may contribute to loss of mobility or conspire to exacerbate the effects (27). Bliss (1992) (28) maintains that healthy people, including those with reduced mobility, do not normally develop PU, but that those same patients may become acutely vulnerable following the onset of an intercurrent illness or trauma.

Intrinsic risk factors (factors inherent in individual patients) include old age 3, 4, 9, 29, incontinence (30), malnutrition 31, 32, reduced food intake, female gender, ASA‐status and NYHA status (29) and anaemia (33) reducing oxygen delivery to tissue which may already be ischaemic (34). Incontinence and moisture might predispose for PU caused by shearing trauma. The moist skin causes the body to adhere to the mattress, thus increasing shearing forces (35). Friction is also exacerbated by the presence of moisture 35, 36. Excessive skin moisture, particularly incontinence has thus been reported to be one risk factor for PU development 37, 38. Moisture lesions have recently been discussed to be distinguished from PU (39). Dry skin, on the other hand, also increases the risk for injury because the elasticity of the skin is decreased (40). The effect of dehydration on tissue viability has been discussed by Livesley (1990) (34). General comorbidity (41) diastolic blood pressure ≤60 mm (42) and confusion (25) have also been reported to be significant risk factors for PU in patients with hip fracture.

Multiple pathologies

The aetiology of PU, therefore, appears to be multifactorial and often a result of multiple pathologies. This might explain the strong relationship noted already between old age and the formation of PU (27).

Versluysen (1986) (6) found that 91of 100 patients who were admitted with an expected diagnosis of fractured neck of femur shared a total of 315 separate diagnoses of concurrent illnesses of different body systems.

Extrinsic risk factors include shearing forces (43) and friction (43). Intensive care (3) and hypothermia (26) have also been reported to contribute to development of PU in hip fracture patients.

Time‐related risk factors reported for surgical patients are long waiting time before surgery (3, 33, 44) and length of surgery (9). Surgery per se (29) and length of surgery ≥4 hours (45) have been reported to be significant risk factors for PU development. A direct association between length of surgery and PU was not confirmed by Kemp et al. (46). The potential effects of shorter surgery have not been studied.

Risk assessment instruments

Prediction of potential complications to surgery in order to avoid them is an important quality indicator. The possibility to predict PU should have positive consequences both for the individual at risk and for the society.

There are several risk assessment instruments for PU available, but none is specifically designed for patients undergoing surgery 47, 48. The Braden risk assessment instrument has been proven to have a high reliability and validity (49) and to be the most reliable of those reported in the literature (50), even if its predictive value has been disputed in patients in intensive care, and when used by non trained staff (51). In one study, where medical record abstraction was performed in 545 patients with hip fracture, risk assessment was reported to be under‐utilised, inaccurate and unrelated to PU development in hip fracture patients (24). Recent studies by Houwing et al. (2004) (9) and Schoonhoven et al. (2002) (45) have also disputed the value of risk assessment tools for prediction of PU for any patient. One critical point that has been launched is the absence of simultaneous recordings of preventive actions taken (52). It can be hypothesised, however, that a validated risk assessment tool like the widely used Braden scale assists in identifying patients at some risk of developing PU during a care episode related to hip fracture surgery.

The aims of the present European, prospective cohort study were

• • to investigate prevalence and incidence of PU upon arrival and at discharge from hospital, and to identify potential intrinsic and extrinsic risk factors for PU development in patients admitted for hip fracture surgery.
• • to illuminate potential differences in PU prevalence, patient logistics and care of potential importance for PU between Northern and Southern Europe.

Patients and methods

The study was designed as a prospective cohort study with inclusion of 20 consecutive patients with a radiologically verified diagnosis of cervical or trochanteric hip fracture per centre. Each country had an experienced study coordinator who was a trustee member of the European Pressure Ulcer Advisory Panel (EPUAP) and who was responsible for the selection of centres and education of the local investigators and staff.

Study protocol

The study protocol agreed upon was designed in three main sections. Section A was aimed at collecting patient‐ and care‐related data at the Accident and Emergency (A&E) Department. Section B comprised questions related to perioperative care and in section C data regarding postoperative care were recorded.

The patients were followed up until discharge or for 7 days, whichever was shorter.

The patients’ skin was daily inspected in specified locations (occiput, scapulae, hips, sacrum, ischium, elbows, heels, back of calves and thighs and ankles) and documented on an anatomical drawing. Classification of PU was standardised and a ‘pressure ulcer card’ with colour pictures guiding the investigators to the correct classification.

Inclusion

One study coordinator (trustee of EPUAP) per country selected Departments of Orthopaedic Surgery willing to participate in the study.

Consecutive patients with radiologically verified cervical or trochanteric hip fractures who after verbal and written information gave their consent to participate were included. In cases where the patient was confused, consent was given by their next of kin.

Exclusion

Patients with multitrauma were excluded from the study.

Definition

A PU was defined as damage to skin and underlying tissues caused by pressure, shear or friction, or a combination of these factors (53).

Pressure ulcer prevalence was defined as PU upon arrival. Pressure ulcers ‘at discharge’ included patients with PU present upon arrival and persisting throughout the care episode as well as PU developed during the care episode (n = 37) and persisting until discharge. Pressure ulcer incidence represents only PU that developed during the care episode and were present at discharge and excludes PU prevalent upon arrival.

Classification of PU

The ulcers classified into four grades according to the EPUAP (53) classification guide.

• • Grade 1, Non blanching erythema of intact skin.
• • Grade 2, Partial‐thickness skin loss involving epidermis, dermis or both.
• • Grade 3, Full‐thickness skin loss involving damage to or necrosis of subcutaneous tissue that may extend down to but not through underlying fascia.
• • Grade 4, Full‐thickness skin loss with extensive destruction, tissue necrosis or damage to muscle, bone or supporting structures.

Sweden, Finland, United Kingdom is abbreviated to North and Spain, Italy, Portugal is abbreviated to South.

Risk assessment

The Braden risk assessment instrument was used. The instrument was translated and re‐translated into the local languages. The Braden scale consists of six domains, which are valued and given specific subscores, which are summarised to one total score.

Risk assessment according to the Braden scale was performed within the first few hours after arrival to the A&E. A cut‐off score of 16 was used (51). Maximum score is 23 (healthy persons).

Patient‐related variables

Intrinsic factors not included in the Braden instrument are documented in Table 1.

Table 1.

Other intrinsic factors documented

Sex and age

Body mass index

Hunger, thirst and dehydration

Comorbidity

Haemoglobin

Blood pressure

Skin condition

The condition of the skin of all patients arriving at the A&E was inspected for signs of moisture and PU during the first two post‐admittance hours. All body locations with particular attention to bony prominences were inspected according to a graphic figure displaying anatomical front and back projections.

Extrinsic/logistic factors recorded were as follows:

• • Site of the accident,
• • Time between arrival to A&E and surgery (hours),
• • Thickness of trolley mattresses at A&E,
• • Traction,
• • Type of anaesthesia,
• • Duration of surgery (hours) and
• • Intraoperative warming

Statistical analysis

Statistical analysis was performed with the help of SPSS (SPSS, Kista, Sweden) using group statistics, independent samples test, chi‐square, t‐test, Mann–Whitney U‐test, Wilcoxon, Z‐test, asym. sig. (two‐tailed), exact sig. (one‐tailed) and cross tab.

Results

Demographic data

The number of patients included were n = 635. The number of patients included per country are itemised in Table 2.

Table 2.

Included patients (n) and PU prevalence upon arrival and at discharge per country

	n	Prevalence (%)	Discharge (%)
North
Finland	300	7	26
Sweden	107	13	21
Great Britain	61	10	18
Total	468		26
South
Portugal	24	4	4
Italy	41	0	16
Spain	104	12	16
Total			16

Prevalence of PU upon arrival

Sixty‐one patients of 599 (10%) had PU upon admittance to A&E [47/441 (10·4%) in North and 14/158 (8·9%)] in South. Fifty‐two PU (87%) were grade 1 ulcers, seven were grade 2 and two were grade 3. There was no grade 4 PU upon arrival to A&E. The most frequent locations were sacrum, heels and ischium.

Pressure ulcers at discharge and incidence

The PU of 21 patients who had PU documented upon arrival healed during the care episode, nine (43%) healed during the first day.

The number of patients with PU at discharge was 131/609 (22%). Pressure ulcers at discharge were more common in patients in the North (26%) than in the South (17%) (P = 0·058). Because 37 patients with PU upon admittance did not heal during the care episode, the incidence was 16% (97/614). Most ulcers were grade 1, and there was no grade 4 PU at discharge. Patient inclusion per country and PU prevalence upon arrival and at discharge are shown in Table 2.

Sex and age

Sex	%
Male	25
Female	75

Pressure ulcers at discharge were equally common in men (21%) and women (22%).

The median age of the patients was 80 years (range 37–99 years). In patients ≤70 years (n = 93), 12 (12%) had PU at discharge. Of patients ≥71 years (n = 499), 121 patients (23%) had PU at discharge (P = ·002).

Incidence of PU in relation to age was 8% for patients ≤70 years (n = 93) and 19% for patients ≥71 years(P < ·000).

Of the 37 patients with PU both upon arrival and at discharge, the median age was 83 years.

Types of fractures versus incidence of PU

Cervical fractures were more frequent (n = 353/550) than trochanteric (n = 197/550) (P = ·051). Cervical fractures were more common in the North (65%) than in the South (41%). Not specified (n = 52).

There was a weak correlation between cervical (P = ·433) and trochanteric (P = ·054) fractures and PU. Patients with trochanteric fractures had a slightly higher age (mean 81 years, median 82 years) than patients with cervical fractures (mean 78 years, median 80 years).

Site of fall accident

Eighty per cent of the patients had fallen and fractured their hips indoors both in the North and in the South.

Inspection routines for PU at A&E (n = 476)

In the North 1% and in the South 8% of the patients were routinely assessed for PU upon arrival prior to the study.

Pain upon arrival

Pain was routinely assessed upon arrival to A&E in 69% of the patients in the North and 33% of the patients in the South (P = 0·000) prior to the study.

Pain was reported by 473/568 patients (83%). Pain intensity upon arrival was registered in 406 patients (VAS 1–10, mean value 5·3, median 6).

There was no statistically significant correlation between pain upon arrival and PU incidence (P = 0·579) (Pearson’s chi‐square).

Smoking versus PU at discharge

Seventy‐seven of the patients were smokers. Eighteen of the smokers (23%) had PU at discharge. There was no significant correlation between smoking and PU at discharge (P = 0·762) or incidence of PU (P = 0·866).

Blood pressure

There was no significant correlation between systolic blood pressure ≤100 mmHg upon arrival (P = 0·220) or low diastolic blood pressure (<80 mmHg) upon arrival and PU at discharge (P = 0·205). In 50 patients with diastolic blood pressure ≤60 mmHg, 7 had PU (P = 0·694).

Haemoglobin (Hb)

One hundred and ninety‐nine patients had a Hb ≤10 g/dl. There was no statistical correlation between Hb ≤10 g/dl and PU incidence (P = 0·150).

Hunger, thirst and dehydration upon arrival

One hundred and fourteen patients of 611 (40%) were hungry upon arrival

Dehydration upon arrival (dry lips, thirst and skinfold test) was significantly more common in North (85%) compared with South (15%) (P = 0·000) and significantly correlated to PU at discharge (P = 0·005). Thirst and dehydration was not statistically correlated to PU incidence (P = 0·493 and P = 0·216), but of the 37 patients with PU at discharge, 47% were dehydrated upon arrival (P = 0·005). Symptoms at A&E and symptoms related to PU at discharge are shown in Table 3.

Table 3.

Symptoms at Accident and Emergency Department (any symptom, n = 337) and symptoms versus pressure ulcers (PU) at discharge (n = 134) Pearson’s chi‐square test

Symptom %	Total n (%)	North n %	South n %	P value North/South	Versus PU P value
Hunger	114/611 (40)	88/443 (18*)	26/168 (33)	0·214	0·410
Thirst	283/606	241/445	42/161	0·000	0·577
Dehydration	144/616	123/454	21/162	0·000	0·005

Body mass index (n = 471)

The body mass index (BMI) had both mean and median of 24 (range 14–47). For the 37 patients with PU both upon arrival and at discharge, the mean was 22·21 (median = 22·56). A higher percentage of patients who were underweight (30%) compared with patients with a normal weight (24%) or overweight (19%) had a PU at discharge. There was no statistical correlation between low or high BMI and PU at discharge (P = 0·142) (Table 4) or between BMI and incidence of PU (P = 0·981).

Table 4.

Body mass index (BMI) (n = 471)

BMI	North (%)	South (%)	Total (%)	P value North/South	P value versus PU
≤20	19	15	18	0·058	0·142
21–24	38	30	36
≥25	43	55	46

Comorbid conditions (n = 452, diabetes n = 537)

Spain reported diabetes as the only comorbid condition.

Thus, concomitant diseases were reported in a cohort of 452 patients and diabetes in 537 patients. Cardiovascular disease was more common in South (Spain no report), 32/61 (52%) compared with North, 90/391 (23%). Diabetes was more frequently reported by the South, 19/146 (13%) compared with the North, 17/391 (4%). Cardiovascular disease was most frequently (78%) correlated to PU at discharge in Sweden (P = 0·00) and United Kingdom (UK) (P = 0·02) (few observations). Diabetes was significantly correlated to PU at discharge (P = 0·005). Comorbid conditions versus PU at discharge are shown in Table 5.

Table 5.

Comorbid conditions versus pressure ulcers (PU) at discharge

Disease/abnormal values	Total n (%)	P value versus PU	Incidence (P)	PU arr + disch (n = 37) n
Gastrointestinal	59/624	0·085	0·232	6
Cardiovascular	323/605 (78)	0·168	0·044	17
Diabetes 1 and 2	90/609 (30)	0·005	0·053	10
Urological	54/605 (13)	0·697	0·561	3
Malignancy	52/606 (16)	0·101	0·156	4
Pulmonary	83/601 (24)	0·083	0·006	6

Braden risk assessment score compared with PU at discharge and with incidence of PU

Risk assessment was routinely performed prior to the study in 34% in the South and 13% in the North.

There was a significant correlation between a low Braden score (≤16) and PU at discharge (P = 0·05) (Mann–Whitney U‐test). When the activity variable was removed, the P value was P = 0·04. There were major inconsistencies in the filling in of the activity and mobility variables. In many cases, the pre‐fracture status was marked. The significance of total Braden score versus PU incidence when activity and mobility variables were excluded was P = 0·036.

Braden risk subscores and PU are shown in Table 6 and Figure 1A–C).

Table 6.

Braden risk subscores versus patients with pressure ulcers (PU) at discharge (n = 131) and versus incidence (n = 97)

Braden subscore	Braden subscore versus PU at discharge (n = 131), P value	Braden subscores versus incidence of PU (n = 97), P value
Activity	0·14	0·217
Mobility	0·21	0·704
Moist skin	0·00	0·004
Friction	0·02	0·031
Nutrition	0·02	0·069
Sensory perception	0·04	0·019

Figure 1.

(A) Braden scores for patients without pressure ulcers (PU) at discharge. (B) Braden scores for patients with PU at discharge. (C) Braden scores for patients without PU at arrival but at discharge (n = 31), six values missing.

Care‐related factors

Trolleys and thickness of mattresses at A&E Department

Ninety‐five per cent of the patients in the North and 80% in the South were placed on a trolley upon arrival. The thickness of mattresses ranged between a mean of 6 cm (Italy) and a mean of 10 cm (Portugal).

There was no significant correlation between thickness of A&E trolley mattress and PU at discharge (P = 0·14).

Time between arrival and start of surgery

Waiting time for surgery had a mean of 55 hours and median of 37 hours (range 58 minutes to 333 hours). In the North, the median time was 48 hours and in the South 90 hours. There was no significant correlation between waiting time to surgery and PU at discharge (P = 0·335) in the total population. Time to surgery for the respective country is shown in Figure 2A and time to surgery and PU at discharge in Figure 2B. The incidence of PU in the hospital with the shortest waiting time (mean 16 hours) versus the incidence in the hospital with the longest waiting time (mean 153 hours) differed slightly, but was not statistically significant (P = 0·075).

Figure 2.

Time between arrival to Accident and Emergency (A&E) Department and start of surgery, all countries. (B) Time between admission to A&E and start of surgery versus patients without pressure ulcers (PU) (no) and with PU (yes) at discharge.

Traction (n = 591)

Traction was used in 6% of the patients in the North and in 74% in the South. The use of traction and PU at discharge or incidence of PU was P = 0·501 and P = 0·511.

Anaesthesia (n = 593)

Spinal anaesthesia was used for 89% of the patients in the North and for 86% in the South. There was no significant correlation between type of anaesthesia and PU at discharge or incidence of PU (P = 0·964, P = 1·000).

Perioperative warming (n = 402)

Warming of operation table and/or warmed infusion was used in a total of 191/402 patients (48%), 62% in the North and 4% in the South. There was no significant correlation between warming and PU at discharge (P = 0·536) or incidence of PU (P = 0·891).

Duration of surgery (n = 587)

Duration of surgery had a mean of 80 minutes and median of 74 minutes. Statistically shorter duration was reported in the North (mean 74 minutes, median 65 minutes), compared with the South (mean = 97 minutes, median 90 minutes) (P < 0·000). Duration of surgery for cervical fractures had a mean of 100 minutes (median 79 minutes) with hip replacement, and without hip replacement it was 72 minutes (median 60 minutes). For trochanteric fractures with hip replacement, mean time for surgery was 95 minutes (median 90 minutes). There was no significant correlation between duration of surgery and PU at discharge (P = 0·845) or incidence of PU (P = 0·804). There was no significant correlation between hip replacement or not and PU at discharge (P = 0·817).

For patients with surgical procedures lasting ≥120 minutes, 85/575 (15%) had PU. The relationship between lengthy surgery and PU incidence was not significant (P = 0·433).

Postoperative turning schedule

A turning schedule was planned for 19% of the patients in the North and for 52% in the South (P = 0·000). The majority of the patients with a turning schedule were turned every second hour.

Three hundred and twenty‐eight of 483 patients had no turning schedule (68%). Twenty‐three per cent without turning schedule and 24% with turning schedule had PU at discharge (P = 0·85).

Lack of turning schedule in relation to the incidence of PU was not significant (P = 0·486).

Discussion

Methodological considerations

This study is, to our knowledge, the largest prospective study of a cohort of hip fracture patients in Europe. Although a considerable number of investigators have been involved, which might be hazardous to the validity of the study, it was managed in each country by skilled investigators and trustees of EPUAP. One weakness is the uneven number of patients included in the respective country, which was explained by local conditions. Even if the results for these reasons may be interpreted with caution, some general findings of statistical significance were noted.

The patients were followed throughout the hospital episode or until day 7 post‐surgery, and the skin was observed for PU upon arrival and daily until discharge. The number of patients with PU doubled during the hospital stay. Thirty‐seven of the patients admitted with PU had them throughout the care episode. The prevalence of PU upon arrival was 10% compared with 18% reported in a smaller Swedish study of similar design (4). Also the incidence was lower, 16% compared with 55–29% 4, 54, which might partly be explained by the shorter follow up time in the present study. However, most surgery‐related PU have been reported to develop between admission and the fourth day post‐surgery (4), an observation also described by other authors 9, 10. The low incidence is most likely, however, explained by a general improvement of care with a focus on PU prevention since the previous studies. This is, however, not reflected in the pre‐study routines for skin inspection in A&E, which was only performed in 1% in the North and in 8% in the South. The higher prevalence and higher number of PU at discharge and the higher incidence in the North is surprising, especially because patients in the South had significantly longer waiting time to surgery, longer duration of surgery, more trochanteric fractures and more commonly were treated by traction. There were also more diabetic patients in the South. Unlike a previously reported study (29), there was no gender difference in development of PU. Reduced tissue tolerance is believed to explain the development of PU to some degree. Tissue tolerance is an individual response to external trauma and decreases in old age. Therefore, increasing age makes the tissue more vulnerable to pressure. This can be explained by the decrease of elastin in soft tissue and changes in collagen synthesis resulting in decreased mechanical capacity of the tissue. Elderly people also have less subcutaneous tissue; their muscles lose their tone and the regeneration of skin cells is slower (35). Hoshowsky and Schramm (1994) (55) found the risk of developing a PU after surgery tripled in people ≥70 years and older. In the present study, age ≥71 years was significantly associated with PU at discharge (P = 0·002).

The majority of patients both in Northern and Southern Europe had fallen and fractured their hips indoors. Cervical fractures were more common than trochanteric (P = 0·051), especially in the North. Pressure ulcers were more common in patients with trochanteric fractures (P = 0·054), which is probably explained by the higher mean age in this group of patients. It has been hypothesised that pain contributes to development of PU by immobilising the patient and by reducing the capillary flow in the skin. Pain was reported by 83% of the patients upon admittance, but was, however, not significantly associated with PU development. Routine assessment of pain upon arrival was more than twice as common a procedure in the North (69%) than in the South (33%). There was no statistically significant correlation between the patients smoking habits, low diastolic or systolic blood pressure and development of PU. Berlowitz and Wilking (1989) (56) and Bergstrom and Braden (1992) (42) reported an association between the development of PU and a decrease in nutritional intake. One fourth of the patients in the present study reported hunger upon arrival, but there was no significant correlation between hunger and PU at discharge. Dehydration, diagnosed by thirst, dry mouth and skinfold was a significant risk factor (P = 0·005) for PU development. Significantly more patients in the North (P = 0·000) compared with the South were thirsty and dehydrated upon arrival, which might be one explanation of the higher incidence of PU in the North. This hypothesis may also be strengthened by the fact that patients in the North were undergoing surgery significantly earlier than those in the South, giving the South patients more time to adjust to normal hydration before surgery. In this study, unlike the Bergstrom and Braden (1992) (42) study, neither a low nor a high BMI was significantly associated with increased risk of PU development, even if underweight patients more commonly had PU. In the group of patients with persisting PU (n = 37), the BMI was lower, but still within the normal interval. Low Braden nutrition subscore was not statistically significant for PU incidence in this study, but was so for PU at discharge (P = 0·02).

Diseases that have been reported to be associated with an increased risk for PU development are pulmonary diseases, anaemia, diabetes mellitus, spine injury and vascular disease 35, 55, 57, 58. Diabetes was in the present study, a significant risk factor for development of PU (P = 0·005). Pulmonary disease (P = 0·006) and cardiovascular disease (P = 0·044) were also risk factors for PU incidence.

Risk assessment was almost three times more frequently undertaken in the South than in the North (34% versus 13%) prior to the study. Both the total Braden risk assessment score (P = 0·05) and the majority of the subscores were significantly associated with PU development. There seems to have been discussions about the registration of mobility and activity in the Braden scale because in some cases, the pre‐fracture status of the patient was documented and not the present status. This led us to calculate both the total Braden score and the Braden score without activity and mobility subscores. The strongest subscore correlation was found between moist skin and PU development. Moist skin can be because of perspiration and/or incontinence; the latter factor has been well documented to be a risk factor for PU development (35, 36). Moist skin causes the body to adhere to the mattress, thus increasing shearing forces. Defloor et al. (2005) (39) debates whether skin lesions caused by moisture should be called PU or moisture lesions because these ulcers are not caused by ischaemia but by softening of the skin by moisture.

Care‐related factors were of minor importance for development of PU in this study, and thus, traction, thickness of trolley mattresses, time to surgery and duration of surgery, spinal anaesthesia or perioperative warming did not show any significant correlation with PU at discharge. In previous studies, prolonged operations have been shown to be a significant risk factor for PU development (59). However, hip fracture surgery is a fast procedure nowadays (median 74 minutes), and very few patients have had surgery, which has lasted more than 2 hours.

Hypothermia has been reported to be a potential risk factor for PU development, and perioperative warming has, in one study in hip fracture patients, been showed to reduce the risk of PU (26). Perioperative warming, however, increases the oxygen demand of the tissue and might in fact increase the risk of PU. In surgery ≥10 hours, Grous et al. (1997) (59) found that the use of warming blankets was the only significant risk factor for PU development. In the present study, no such correlation was shown; in fact, in Southern countries where perioperative warming was rare, PU at discharge were less frequent.

Waiting time to surgery has been reported to influence the development of PU. Because the waiting time till surgery was significantly longer in the South, this might have been a significant risk factor but, in fact, fewer patients in the South had PU than those in the North, and there was no statistical significance of long waiting time for development of PU (P = 0·335). However, the hospital with the shortest waiting time had fewer PU than the hospital with the longest waiting time (P = 0·075), indicating that in a bigger study, waiting time might be of significant importance. It might also be so that the hospital with short waiting time has been particularly dedicated to reduce also other risks. In one study from that hospital, waiting time to surgery for hip fracture patients was systematically and substantially shortened, which reduced the number of patients with PU at discharge from 19% in 1998 to 4·5% in 2000 (60). In that study, the number of patients who got pain relief within 1 hour post‐admittance was also doubled during the study period 1998–2000.

Low general prevalence of PU in the South has been reported by other authors (61) and confirmed in the present prospective study.

Preoperative risk assessment was not frequently a routine neither in the South or in the North and its value has been disputed (45). However, bearing in mind the findings of the present study, PU in hip fracture patients can be accurately predicted using the Braden scale, especially if the activity and mobility subscores are removed or clarified. It should be recognised, however, that risk calculators were never designed to replace clinical judgement but rather to assist in decisions, which channel resources appropriately (27). Even if the validation of PU risk assessment scales remains a topic of considerable debate and uncertainty, and bearing in mind the recommendation that they should be evaluated in combination with preventive measures (52), it is likely that they contribute to increased awareness of some risk factors, which, if not observed, can lead to development of avoidable PU.

Recommendations included the following:

• • Risk assessment and skin observation should be performed upon arrival to the A&E and daily until discharge with special attention to patients aged ≥71 years.
• • Pay particular attention to significant Braden risk factors such as moist skin, friction, malnutrition and low sensory perception.
• • Observe and correct dehydration.
• • Observe patients with diabetes mellitus and cardiovascular and pulmonary diseases who might be at risk of developing PU.