This study is reported according to the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement [20].
Conventional hip fracture care
Patients with suspected hip fracture were admitted via the accident and emergency (A&E) department, Akershus university hospital (AHUS). The patients had to wait for an available examination room, an available physician and a slot in the radiology lab. After x-ray examination the patients were transported back to the A&E department where the admitting physician evaluated the x-rays and finished the work-up before the patient was transported to the orthopaedic ward.
Apart from antithrombotic and perioperative antibiotic prophylaxis, perioperative treatment was not standardized. There was no systematic focus on nutritional status and on prevention and treatment of delirium. The patients were not consistently mobilized on the first postoperative day.
Fast track hip fracture care
Fast track hip fracture care was introduced at AHUS in two steps. From 28 October 2013, guidelines concerning perioperative treatment were taken into practice. Secondly, 27 January 2014, the fast track pathway for hospital admission was implemented.
The ambulance personnel initiate first line treatment (intravenous fluids, oxygen, pain relief and electrocardiogram). Upon arrival at the A&E department a trained nurse triages all suspected hip fracture patients using the Manchester triage system [21]. If deemed necessary, additional intravenous opiates are administered in selected cases.
Patients without signs of other, more acute medical conditions (score of 3 or higher) continue in the fast track admission pathway if they fulfil the following criteria: Low energy trauma, hip/groin pain, shortened and/or externally rotated lower extremity and/or unable to bear weight, no sign of other fractures, no sign of neurovascular injury, not previously operated on the hip in question.
From the triage area the patient is brought directly to the radiology lab where fast track hip fracture patients are prioritized after any ongoing procedure.
The radiology technician evaluates the x-ray. If considered to have a hip fracture the patient is transported directly to the orthopaedic ward where a nurse performs standard procedures according to a check-list and gives the patient both oral and written information about hip fractures and the expected course of treatment.
The orthopaedic surgeon re-evaluates the x-ray, writes an admission note, administers a fascia iliaca compartment block and prescribes a set of standard medications, including oral and intravenous fluids and pain medication.
The fast track hip fracture care system includes written guidelines concerning standard blood sampling, premedication, pre- and postoperative pain relief with focus on opiate sparing, pre- and postoperative fluid treatment with focus on short periods of fasting, transfusion-triggers and management of anticoagulants. Patients are mobilized on the first postoperative day. The guidelines also advise on screening for and prevention and treatment of delirium, on screening of nutritional status and on appropriate interventions concerning nutrition.
Data collection
All primary and revision hip fracture operations in Norway should be reported to the Norwegian Hip Fracture Register (NHFR) [22]. This is done prospectively by the surgeon on a 1-page questionnaire which includes information on the type of fracture, American Society of Anaesthesiologists (ASA) score [23], cognitive impairment (possible choices: ‘no’, ‘uncertain’, ‘yes’), type of anaesthesia, type of operation, surgeon’s experience (at least one surgeon with > 3 years of experience in hip fracture surgery) and operating time (time from incision to skin closure).
We obtained the NHFR data for patients operated at AHUS from January 2012 through December 2015. For the same time period, hip fracture patients were identified from the electronic hospital records using the search strings main diagnosis S72.0, S72.1 or S72.2 (ICD-10), in-patient, operated during that hospital admission. The two databases were linked deterministically using the unique 11-digit Norwegian personal identification number. For patients who were only identified in one of the data sources or for whom the records did not match, the electronic hospital records were scrutinized to determine if there had occurred an error in coding or in reporting to the NHFR.
Patients
All patients 18 years of age or older who were operated for a fracture of the proximal femur (femoral neck, trochanteric or subtrochanteric) at a single institution (AHUS) from January 2012 through December 2015 were eligible for inclusion (Fig. 1). AHUS has a catchment area of approximately 500.000 inhabitants.
During the study period 2634 primary hip fracture operations were performed. For patients who sustained two hip fractures during the study period (n = 92), only the first fracture was included in the analysis. Patients with a pathologic fracture were excluded (n = 47). 265 patients were not reported to the NHFR, corresponding to an overall reporting rate from AHUS of 89.4% (87.9% before and 90.8% after the introduction of fast track hip fracture care). The remaining 2230 patients, who were reported to the NHFR, represent the study population. Of these 1090 patients were operated before (conventional treatment group) and 1140 patients were operated after the introduction of fast track hip fracture care (fast track group) with 28 October 2013 as cut-off. The available data do not allow us to determine to what extent the different components of the reported fast track care system were applied to an individual patient. Therefore, the analysis follows the intention to treat principle and hip fracture patients treated from 28 October 2013 and onwards are included in the fast track group, irrespective of length of admission time or other criteria.
Operative treatment
Patients with a femoral neck fracture were treated either with closed reduction and internal fixation with two screws, with a cemented bipolar hemiarthroplasty or with a cemented total hip arthroplasty, both with a taper slip stem using a direct lateral approach. Trochanteric fractures were treated with a sliding hip screw and subtrochanteric fractures with an intramedullary nail. Surgical treatment guidelines did not differ before and after the introduction of fast track hip fracture care.
Outcome measures
Mortality
Mortality data from the Central Population Register are routinely imported to the electronic hospital records and a last up-date of the database was performed on 14 September 2017 to allow for a delay in registration. There was no loss to follow-up. 30-day, 90-day and 1-year mortality were calculated from time of arrival at the hospital. Survival was censored at 1 year.
Reoperations
Data on reoperations were obtained from the NHFR. Time to event was calculated from the index operation. In 2013 and 2014 the reporting rate from our institution was 66% for reoperations after osteosynthesis and 81.6% for reoperations after hemiarthroplasty [24].
Surgical site infection
The Department of Microbiology and Infection Control, AHUS, surveys surgical site infections after hemiarthroplasty and total arthroplasty of the hip under the Norwegian Surveillance System for Antibiotic Use and Hospital-Acquired Infections (NOIS) [25] with 30-day and one-year follow-up. Sliding hip screws are not monitored by NOIS, but the Department of Microbiology and Infection Control, AHUS, surveys surgical site infection also in these patients with a 30-day follow-up using the same criteria. The completeness of the 30-day follow-up ranged from 97.7 to 99.6% per calendar year. The completeness of the 1-year follow-up ranged from 98.7 to 99.5% for hemiarthroplasties and from 87.6 to 91.9% for total hip arthroplasties.
Internal fixations of femoral neck fractures and intramedullary nails are not systematically surveyed for surgical site infection.
Time to event was calculated from the time of the index operation.
Composite 30-day outcome
A patient was considered to have had this (negative) outcome if any of the following had occurred: death within 30-days from arrival at the hospital, reoperation or infection within 30-days from the index operation.
Readmission
Readmission was defined as any cause, non-elective readmission within 30 days after discharge from the index admission. These data were extracted from the electronic hospital records.
Admission time, time to surgery and length of hospital stay
Admission time (time from arrival at the hospital to arrival at the orthopaedic ward), time to surgery (time from arrival at the hospital to skin incision) and length of hospital stay were extracted from the electronic hospital records.
Statistical analysis
A sample size calculation based on a reduction of the 30-day mortality rate after hip fracture from 10.7% (AHUS in 2011 [26]) to 6.8% (hospital with lowest mortality rate in Norway in 2011 [26]), 85% power and a level of significance of 0.05 yielded a total sample size of approximately 1800 patients [27].
Fisher’s exact test was used for unadjusted comparisons of proportions, while the Chi square test was used for unadjusted comparisons of ordinal and nominal distributions. Student’s T test was used for unadjusted comparisons of continuous variables. However, comparisons of admission times, time to surgery and length of hospital stay were made with non-parametric Mann-Whitney U tests, rather than t-tests, due to the skewed distributions of these variables.
Logistic regression was used to analyze the effects of different predictors on the binary outcomes of mortality at 30 days, 90 days and 1 year follow-up, reoperation and surgical site infections at 30 days and 1 year follow-up, as well as 30-day readmission and the composite 30-day outcome. The main predictor of interest in these models was the conventional treatment/fast track care indicator. Other variables were included as confounders if they showed statistical significance at the 0.05 level, except for patient age and gender, which were always included. All municipalities and Oslo districts belonging to AHUS’ catchment area were included in the analysis with a distinct identifier while patients from outside our hospital’s catchment area were coded as one group. This variable was considered as a random effect in the models, but turned out not to have a significant effect. Reducing time to surgery is one of the intended effects of fast track care. Therefore, the logistic regression model analyzing the effect of fast track care on mortality was run with and without including time to surgery as an independent predictor. This did not relevantly change the result for the effect of fast track care on mortality.
Survival analysis by Cox regression was considered for the binary outcomes, since these were all associated with event times. However, there were problems with the assumptions of proportional hazards, measured by Schoenfeld residuals ph-test. Concerning mortality there was no loss to follow-up, so the logistic regression models’ inability to handle right-censoring was not an issue. Also, the standard quality indicators of hip surgery are defined as the number of adverse outcomes after 30 days, 90 days and 1 year, which is in line with logistic regression. Logistic regression was therefore chosen.
A post hoc power analysis was performed using standard normal distribution approximation.
Subgroup analyses
Patients were divided into two subgroups according to their comorbidity (ASA score). Frailer patients were defined by an ASA score ≥ 3.