Introduction

Hip fractures are a serious and common health care problem in the elderly, with an incidence ranging from 20 to 921 per 100,000 population worldwide [1, 2]. Geriatric hip fractures are most frequently caused by osteoporosis and are associated with an increased mortality and morbidity in the short term and long term [3,4,5,6,7,8].

Several guidelines for the prevention and treatment of hip fractures are available and multiple recommendations exist on the multidisciplinary approach for the treatment of geriatric hip fracture patients [9,10,11,12,13]. In some countries, adherence rates to these quality standards, described in the guidelines, are routinely collected in national databases. In a landmark US study, the percentage of recommended care received by patients with a hip fracture was 22.8%, which was the second lowest for 25 clinical conditions [14]. The data of the National Hip Fracture Database (NHFD) in England, Wales, and Northern Ireland show large variation between hospitals, but some practices have improved over the years [15]. The DUQuE project (Deeping our understanding of quality improvement in Europe) aimed to study the effectiveness of quality improvement systems. In this European study, 5.9% of the patients received all recommended care and the overall percentage of recommended care received was 60% [16, 17].

First, to date, no international study has focused on all care aspects which hip fracture patients should receive or on the adherence to guidelines during in-hospital treatment. These are the care aspects regarding preoperative, perioperative, and postoperative management. Second, healthcare managers need more guidance in identifying priority areas for improvement. This would allow them to bridge the gap between guidelines and the interventions that are actually performed in daily practice. Shortcomings can be visualized by an importance–performance analysis, assisting hospital managers to prioritize care interventions for which performance rates need to be improved. Our study describes a multidisciplinary audit of the treatment of geriatric patients with a hip fracture. We propose an importance–performance analysis that combines importance rates (expert consensus on the importance of the clinical activities on patient outcomes) and performance rates (adherence to guidelines). Four quadrants result from this combination and serve as a method for prioritizing care interventions. Such importance–performance analysis was initially introduced for the development of marketing strategies [18]. While its value is also recognized in healthcare, actual applications such as the one presented in this study are limited [19, 20]. The aim of this study is thus threefold: (1) to describe the level of expert consensus for the in-hospital treatment of geriatric patients with a proximal femur fracture, (2) to describe the adherence to guidelines for these patients, and (3) to conduct an importance–performance analysis for prioritizing interventions to improve the in-hospital management of these patients.

Methods

A secondary data analysis was conducted from the data collected in the EQCP study (European Quality of Care Pathways Study). The EQCP study is an international cluster randomized-controlled trial performed in Belgium, Italy, and Portugal. The aim of the EQCP study is the evaluation of the impact of care pathways on team, process, and outcome indicators. Patients consecutively admitted for proximal femur fracture were included if (1) they provided written informed consent; (2) they were minimum 65 years; (3) they had a closed fracture and were eligible for surgical intervention; (4) they had an American Society of Anesthesiology (ASA) score of 1, 2, or 3; and (5) they were able to communicate in the native language. Patients were excluded if they had severe dementia based on DSM IV criteria, pathological fracture, or peri-prosthetic fracture [21]. Hospitals were randomized in an intervention and in a control group. First, the adherence to guidelines was measured before the implementation of a care pathway (pretest data). Second, the adherence to guidelines was measured after implementation of a care pathway (posttest data) and in the control group [21]. For the current study, the pretest data and the data in the control group were used as these reflect usual care. Patients were included in the EQCP study between April 2009 and May 2014. In Belgium, 125 patients in 12 hospitals were included between April 2009 and August 2009 (pretest data) and 114 patients in 6 hospitals were included between October 2010 and January 2012 (control group). In Italy, 90 patients in 5 hospitals were included between May 2011 and December 2011 (pretest data) and 38 patients in 2 hospitals were included between February 2013 and May 2014 (control group). In Portugal, 112 patients in 6 hospitals were included between January 2011 and November 2011 (pretest data) and 61 patients in 3 hospitals were included between February 2013 and August 2013 (control group). Overall 540 patients in 34 hospitals were included.

Level of expert consensus (Importance analysis)

Before the start of the data collection, process and outcome indicators were identified based on the eight-step method to build the clinical content of an evidence-based care pathway [22, 23]. The flow diagram of the eight-step method is shown in Fig. 1. Twenty-two process indicators were identified and classified in the following core processes: preoperative management (7 process indicators), perioperative management (2 process indicators), and postoperative management (13 process indicators). The level of expert consensus, based on an international Delphi study [22, 23], was used to define the importance rate. Each expert scored the different care activities, as mentioned in the different guidelines, based on their importance to be followed up for the in-hospital management of patients admitted with PFF. The experts involved in the Delphi panel had different nationalities and backgrounds. For this Delphi study, two rounds were performed and the level of expert consensus was defined as the percentage of participants who scored the indicator 5 or 6 on a 6-point Likert scale.

Fig. 1
figure 1

Overview of the eight-step methodology for inclusion of quality indicators. The left side contains the theoretical eight-step framework; the right side contains the application of the eight-step methodology to the current study

Full size image

Adherence to guidelines (Performance analysis)

A retrospective patient record analysis was performed by an external researcher to evaluate the performance rate regarding the 22 process indicators [21]. Adherence to guidelines (performance analysis) was measured as a proportion and has as numerator the number of individual process indicators adhered to and as denominator the number of relevant process indicators. The denominator can be different for each patient as not all process indicators are applicable to each patient.

Importance–performance analysis

Indicators with a level of expert consensus above 75% were classified as high level of agreement indicators (importance rate) [22, 23]. As there is no established benchmark, indicators with an adherence to guidelines above 80% were classified as highly performed (performance rate) [24]. The combination of importance rate and performance rate is visualized through an importance–performance grid. Four quadrants are identified [18] (Fig. 2). The first quadrant contains the “appropriate use care interventions” with high importance and high performance rates. The second quadrant contains the “underuse care interventions” with high importance rates and low performance rates. These are the interventions to prioritize to improve the quality of care. The third quadrant and fourth quadrant contain the low priority care interventions [18, 25].

Fig. 2
figure 2

Importance–performance analysis. The numbers in the figure correspond to the numbers mentioned in Table 1. Dark gray: preoperative management process indicators; light gray: peri and postoperative management process indicators. The cut-point for importance (75%) is derived from an international Delphi study [22, 23]; and as there is no established benchmark, the cut-point for performance was set at 80% [24]. The size of the circles indicates the amount of between-hospital variation. The larger the diameter, the larger the IQR (Q3–Q1)

Full size image
Table 1 Process indicators at patient and hospital level
Full size table

Statistical analysis

Descriptive statistics were performed on patients and hospitals. The importance rates and performance rates were presented by percentages. For each hospital, the performance at hospital level was calculated by aggregating the patient data of that hospital. Variation between hospitals, for each indicator, was described by the median, first quartile, 3rd quartile, and the interquartile range (difference between third quartile and first quartile). Variation in total adherence to guidelines across and within hospitals is shown by a boxplot. Analyses at patient and hospital level were conducted in R using packages easyGgplot2 and ggplot2.

Compliance with ethical standards

The study was registered at ClinicalTrials.gov (NCT00962910) and ethical approval was received in each country (Identifier Belgium: B32220096038, Identifier Italy: 625, 21/07/2011, Identifier Portugal: 6605/2011), and from all hospitals. Written informed consent was obtained from all patients [21].

Results

Importance rate

The level of expert consensus for 22 process indicators is shown in Table 1. Consensus between experts varied from 24.1% for assessment of prefracture falls to 96.3% for extra support arranged in patients who need more support when going home. Level of expert consensus was high (above 75%) for 12 of 22 process indicators.

Performance analysis

Patient and hospital characteristics are described in Table 2. Three in four patients were female (74.1%) and were living at home before the fracture (74.8%). Fifty-seven percent of the patients could walk without help before the fracture. Use of vitamin D and calcium before fracture was given in respectively 3.1 and 3.7% of the patients. Forty-six patients (10.1%) were referred to osteoporosis clinic. Two in five hospitals were teaching hospitals (41.2%) and 17.6% of the hospitals admitted at least 300 patients with proximal femur fracture yearly.

Table 2 Patient and hospital characteristics
Full size table

Table 1 displays the performance rates for the 22 process indicators at patient and hospital level. The adherence to the individual performance rate at patient level varied between 91.7% (adequate administration of analgesia postoperatively) and 0.2% (assessment of postoperative nutritional status). There is large variation in performance rates between patients and the largest variation between the indicators was noticed for postoperative management. The adherence to the individual process indicators at hospital level is shown by the median, first quartile, and third quartile, and varied between 100% (adequate administration of analgesia postoperatively) and 0% (assessment of prefracture falls, performance of antibiotic prophylaxis, assessment of cognitive status at start of mobilization, performance of antithrombotic prophylaxis, assessment of fluid balance, performance of adequate pain assessment postoperatively, and assessment of postoperative nutritional status). The adherence to the received care bundle, containing all the care interventions which a patient should receive, varied from 77.3 to 0%. Patients received 38.7% of the recommended care and only 21.5% (116 out of 540 patients) received 50% or more of the care bundle. Figure 3 describes the variation in adherence to guidelines across hospitals, which ranges from 4.5 to 54.2%. The average interquartile range (IQR), difference between first quartile and third quartile, for the eight highest scoring hospitals was 10%, while this was 12.9% for the eight lowest scoring hospitals.

Fig. 3
figure 3

Variation in the total adherence to guidelines within and between hospitals. The box contains, respectively, the first quartile, median, and 3rd quartile. The outliers are visualized by black dots

Full size image

Importance–performance analysis

The importance–performance grid is displayed in Fig. 2. In quadrant 1, three appropriate use indicators with both a high importance and high performance rate are described. These are mainly interventions regarding preoperative management. Performance of these indicators should be maintained.

In quadrant 2, nine underuse interventions are described. These interventions have a high level of importance (expert consensus > 75%), but a low performance rate (adherence to guidelines < 80%) and are thus high priority areas for improvement. The four lowest scoring interventions did not attain performance over 20% and are interventions regarding postoperative management.

The bottom quadrant contains ten low priority activities with a low importance and low performance rate (quadrant 3). There are no interventions with a low importance and high performance rate (quadrant 4).

The size of the circles indicates the amount of between-hospital variation (the larger the circle, the larger between-hospital variation; see also right column in Table 1). The highest IQR (79.4) was shown for the indicator assessment of prefracture mobility status and the lowest IQR (1.2) was shown for assessment of fluid balance. The IQRs for process indicators with a performance rate below 25% or above 75% are lower compared to the IQRs for process indicators with a performance rate between 25 and 75%. Two indicators had an IQR of zero. These indicators were performance of adequate analgesia postoperatively (process indicator 21) and assessment of postoperative nutritional status (process indicator 22) (Table 1; Fig. 2).

Discussion

This is the first study auditing the complete in-hospital care for patients with a geriatric hip fracture. Two findings emerge from this study. First, despite our expectation that the 12 indicators rated by experts as highly important would have a high performance rate, only three of them were performed for more than 80% of the patients. Furthermore, there remains a large variation to the total adherence to guidelines within and between hospitals. Second, the proposed importance–performance analysis can help hospital managers prioritizing and improving the in-hospital care for geriatric hip fracture patients.

The performance rates for geriatric patients with a hip fracture are in line with the previous studies. The proportion of recommended care received in our study (38.7%) is higher than in the study performed in the US (22.8%) [14] but is lower than the results achieved in the DUQuE project (60%) [16, 16]. In our study, no patient received the full care bundle, while, in the DUQuE project, 5.9% of the patients received all care interventions. A likely explanation is the inclusion of 22 process indicators in the current study compared to only four in the DUQuE project [16, 17]. Due to a lack of consistency in the measured process indicators, it is difficult to compare the results between the different studies. In the US study, for example, nine indicators were derived from the Research and Development’s (RAND) Quality Assessment Tool System [14], while in the DUQuE project, this was a combination of the RAND’s Quality Assessment Tool System (2 indicators), Danish Clinical Registries (1 indicator) and Health Care Quality Indicator programme of the Organisation for Economic Coorperation and Development (OECD) (1 indicator) [16, 17].

The results of our study confirm that geriatric patients with a hip fracture show a low overall compliance to the evidence-based organization of care [26]. There is an urgent need to improve the care for these patients. Several quality improvement initiatives at national and local levels have been set up and have been described in the literature [27,28,29]. An example initiative at national level is the implementation of a clinician led audit initiative [27, 29]. To achieve the best results of audits and feedback, feedback should be provided as closely as possible to the actual patient care given. Therefore, continuous team feedback is needed [30]. Quality interventions at department level are positively related to patient-related process and outcome parameters. Quality improvement systems in the whole organization are positively related to departmental quality interventions [30, 31]. Focusing on all the process indicators at once can be burdensome and complex. Therefore, hospitals should set priorities to improve the care for patients with proximal femur fracture, which can be identified by an importance–performance analysis. In this study, the importance analysis was visualized by the level of evidence based on the results of an international Delphi study. A Delphi study is not always available and the grade of recommendations or level of evidence, mentioned in guidelines, can be used as importance rate. This was already used for surgical patients with colorectal cancer [32]. Based on this importance–performance analysis, the priorities for hospitals treating geriatric hip fracture patients are: performance of hospital visits by social worker during hospitalization, assessment of cognitive status at start of mobilization, performance of antithrombotic prophylaxis, and assessment of fluid balance. Quality improvement initiatives should thus first focus on the process indicators with a low performance rate and with a high importance rate. Some of these process indicators fit in the main challenges for hip fracture management described in the literature, which are early surgery, dedicated medical care, and early rehabilitation after operative hip fracture treatment. These interventions can improve outcomes and reduce mortality [26, 33,34,35,36,37,38]. Interventions with less variation between hospitals are mainly interventions with a performance rate below 25% or above 75%. The indicators with a performance rate above 75% are indicators regarding adequate pain assessment preoperatively, assessment of hemoglobin preoperatively, and adequate administration of analgesia postoperatively, and are thus mainly interventions closely related to surgery. In total, eight care activities have a high-level evidence, based on the American Academy of Orthopaedic Surgeons. Out of these, five care activities are mainly closely related to surgery and are preoperative regional analgesia, anesthesia, displaced femoral neck fractures, subtrochanteric or reverse obliquity fractures, and transfusion threshold. The other care activities with high-level evidence are intensive physical therapy, interdisciplinary care program and postoperative multimodal analgesia and are more related to the postoperative management [39]. Indicators with a performance rate below 25% are indicators regarding assessment of prefracture falls, visit by social worker during hospitalization, assessment of cognitive status at start of mobilization, correct performance of antithrombotic prophylaxis, assessment of fluid balance, adequate pain assessment postoperatively, and assessment of postoperative nutritional status. These indicators are more linked to preoperative and postoperative management and are more diverse.

More than 50% of the patients with a hip fracture are older than 86 years [40] and are suffering from multiple comorbidities [41]. Amongst them, osteoporosis is in important problem. In this study, we looked at the care patients with proximal femur fracture received in general and less at the subgroup of patients with osteoporosis. It is remarkable that only a small amount of patients were referred to an osteoporosis clinic. This can be due to the variability in treatment for these patients between the different hospitals and countries, e.g., internal protocol of referring patients within the hospitals; sometimes, the general practitioner performs the treatment of the underlying osteoporosis. Using FRAX (fracture risk assessment tool), patients can be referred to the osteoporosis clinic more focused [42].

The current study has several strengths. A first strength is the collection of data performed by local study coordinators. Information was collected from the patient records and missing data values were limited, leading to more reliable data and reduced assessment bias. For the data collection, we strictly adhered to the defined protocols and did not exclude patients based on missing data. However, the high percentage of unknown characteristics are a weakness and this is a problem of data recording. For the ASA score for example, it is well known that the anaesthetist calculates this score before starting with the procedure. Nevertheless, it could not be found in the patient records. On the other hand, there was very poor compliance of the hospitals in reporting missing data despite repeated requests of the study investigators. This was, for instance, the case for age of 36 patients. For these patients, it was indicated that inclusion criteria (including age above 65 years) were met, but the exact age was not provided despite repeated requests of the study investigators. A second strength concerns the selection of process indicators. This is based on the results of an international, multidisciplinary Delphi study, resulting in a complete bundle of care for the in-hospital management of patients with proximal femur fracture. Process indicators with a high importance rate and a low performance rate could be used as a basis for quality improvement programs. The care bundle, each patient should receive, contains between 20 and 22 care interventions as the arrangement of extra support is only needed for patients who are going home and active mobilization within 24–48 h can only be performed in patients who could walk before fracture (Table 1).

The main limitation of this study is its convenience sample. We should, therefore, be careful in generalizing the non-adherence prevalence rates to the three countries involved. Another weakness of this study is that we rely heavily on data that are reported in the patient record. Incomplete documentation and missing information can lead to an underestimation of the received care due to the fact that an intervention was performed but not reported in the patient record. Several healthcare providers are assigned to patients and when an action is not mentioned in the record, this can lead to duplication of actions as the other healthcare providers are not aware of this action. Accurate recording of the care process is needed to have a detailed measure of guidelines that were adhered to [43]. A prospective study including complete patient record analysis could give us a complete view of the quality of care of this patient population. However, prospective data collection could have led to more bias. A third weakness concerns data collection based on an international cluster randomized-controlled trial. Therefore pretest data and posttest data of the control group were brought together after the implementation of our intervention leading to older data. Finally, in each hospital, only a low amount of patients was included which can lead to higher variation. This current study used the data of an international cluster randomized-controlled trial (EQCP study); and in this study, it was decided not to increase the number of patients within each hospital as increasing the sample size per cluster increases the power less [44]. On the other hand, each patient admitted in each center should receive all appropriate care activities.

In conclusion, this study shows that patients with a hip fracture are receiving suboptimal care. Actions should be taken to improve adherence to activities of care with a high level of evidence. An importance–performance analysis allows managers and practitioners to identify priority areas for improvements within the local organizational culture.