CLINICAL PRACTICE Short-term mortality in hip fracture patients admitted during weekends and holidays

Transcription

1 British Journal of Anaesthesia 96 (4): doi: /bja/ael012 Advance Access publication January 27, 2006 CLINICAL PRACTICE Short-term mortality in hip fracture patients admitted during weekends and holidays N. B. Foss 1, 2 * and H. Kehlet 3 1 Department of Anaesthesia and 2 Department of Orthopaedic Surgery, Hvidovre University Hospital, Denmark. 3 Department of Surgical Pathophysiology, Juliane Marie Centre 4074, Rigshospitalet, Denmark *Corresponding author: Department of Anaesthesiology, Hvidovre University Hospital, Copenhagen DK-2650, Denmark. nicolai.bang.foss@hh.hosp.dk Background. Acute surgical admission during weekends, with reduced staffing levels, has been associated with increased risk of mortality, but the effect of longer vacation/holiday periods has not been studied. We therefore examined early postoperative mortality in hip fracture patients admitted during weekends and holiday periods, compared with normal weekdays. Methods. Prospective, descriptive study in 600 consecutive hip fracture patients treated with a well-defined multimodal care plan, in a specialized hip fracture unit between September 2002 and July Patients were stratified according to admission on a weekday or during weekends/ holiday periods. Results were analysed with univariate and multivariate analyses. Results. Three hundred and thirty-two patients were admitted during weekdays, 118 during weekends and 150 during holiday periods. Both 5- and 30-day postoperative mortality were significantly higher in patients admitted during holiday periods than during weekends and weekdays, 8.0% vs 2.5% and 1.8%, respectively (P=0.01) and 19.3% vs 12.7% and 11.1%, respectively (P=0.05). In a multivariate analysis, admission during holiday periods was still a significant independent risk factor for both 5-day (4.34, 95% CI ) and 30-day mortality (1.84, 95% CI ). Conclusion. Staff reduction during holiday periods in units that care for acute surgical patients may adversely influence postoperative outcome. This may have important consequences both for outcome analysis of interventions and the planning of resource management in surgical units. Br J Anaesth 2006; 96: Keywords: complications, holiday; complications, mortality; complications, weekend; hip fracture; surgery, fast-track Accepted for publication: December 28, 2005 Hospital staffing may be reduced during weekends and holidays, and admission during weekend periods has been associated with increased risk of mortality in selected medical and surgical diagnoses. 1 3 This pertains, in particular, to deaths occurring within a few days of admission. 1 However, the weekend effect could not be demonstrated in hip fracture patients. 12 No studies have looked at the effect of longer holiday periods, with similar reduction of ward staff, on mortality after acute surgery. We have studied early postoperative mortality in hip fracture patients admitted during the periods of reduced hospital staffing, namely weekends and holiday periods. The setting was a specialized hip fracture unit focusing on optimized multimodal rehabilitation 4 and using a standardized perioperative care plan. 56 Patients and methods Between September 2002 and July 2004, 600 unselected patients with a primary hip fracture admitted to the special hip fracture unit at the Department of Orthopaedics at Hvidovre University Hospital were studied prospectively. This study is part of Hvidovre University Hospitals Hip Fracture Project, which was evaluated by the local ethical Ó The Board of Management and Trustees of the British Journal of Anaesthesia All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 Hip fracture mortality after holiday admission committee, who concluded that no written patient consent was necessary. The study was approved by the Danish data protection agency. Patients were treated according to a standardized care plan 56 including surgery within 24 h from admittance, epidural anaesthesia and perioperative analgesia initiated on admission and continued until the fourth postoperative day. A standardized fluid and transfusion protocol was used. On admission, fluid therapy consisted of rehydration fluid (Na 40 mmol litre 1, K 20 mmol litre 1, glucose 250 mmol litre 1 )20mlkg 1. Intraoperatively an infusion of isotonic saline 5 ml kg 1 h 1 was given, which was supplemented by 500 ml of 6% hydroxyethyl starch 130/0.4 on signs of hypovolaemia. Hypotension was treated by ephedrine 10 mg i.v. and 40 mg i.m. Blood loss was replaced by 6% hydroxyethyl starch 130/0.4, at a rate of 1:1 until haemoglobin fell below the standardized transfusion trigger of 10 g dl 1. In the post anaesthesia care unit (PACU) all patients received 500 ml i.v. (glucose 278 mmol litre 1 ). Postoperative fluid therapy was standardized and i.v. fluids were only given if oral intake was <1500 ml daily, or in the case of hypovolaemia when 500 ml of 6% hydroxyethyl starch 130/0.4 was given. Haemoglobin was measured on admission, immediately postoperative and every morning up to the fifth postoperative day. All patients received 2 litre min 1 of supplemental oxygen whenever supine until the fourth postoperative day. Intensive oral nutritional support was given using three daily supplementary protein drinks. All patients were given prophylactic intraoperative antibiotics, and perioperative low-molecular weight heparin. The patients received a standardized intensive physiotherapy programme starting on the day of operation. Data were gathered on preoperative functional level via the new mobility score (NMS), 7 cognitive function, residential status, fracture type, ASA classification, type of anaesthesia and surgery and length of surgery and intraoperative bleeding. Mortality within 30 days of surgery was assessed using the Danish patient registry. Deaths were classified as occurring within 5 days or within 30 days of surgery, to distinguish between early and later postoperative mortality. During the entire registration period, all days were classified as weekdays, weekends or holidays. An admission was defined as occurring on a weekend if the patient arrived in the emergency room or was referred from another ward between 4 p.m. on Friday and 4 p.m. on Sunday; as the trauma theatre only did surgery until 6 p.m. This ensured that no patient having surgery on a Friday was registered as a weekend admission. A similar definition was applied to holidays, defined as all days where the staffing level of the hip fracture unit was reduced because of planned holidays. The holiday periods were spread throughout the year resulting in reduced staffing for 10 weeks, with 2 weeks during the Christmas/New Year season, 6 weeks in June/July and 1 week each during spring and autumn, respectively, not including bank holidays. Weekends that occurred during holiday periods were considered vacation periods. Admission procedures were uniform regardless of weekends and holidays, and an operating theatre specifically assigned for trauma surgery was available on all days. The hip fracture unit is a dedicated ward with 14 beds that receives patients from the urban part of Copenhagen. The average yearly patient intake is about 320 patients. Admission of hip fracture patients was done jointly by the nurses in the emergency room and the junior orthopaedic resident on call at all times. After admission, the patient had a preoperative assessment performed by the anaesthesia junior resident, who at the same time inserted an epidural catheter for epidural analgesia. This procedure was the same regardless of weekends and holidays and was performed in the PACU. Surgery was performed by different grades of orthopaedic surgeons depending on the complexity of the individual surgical procedure. Anaesthesia was provided by the junior resident of anaesthesiology, using the epidural catheter, along with specially trained anaesthesia nurses, regardless of weekend/holiday periods. After operation, all patients were taken to the PACU, until deemed stable for discharge, regardless of weekend/holiday periods. The staffing level for nurses and healthcare assistants on the 14-bed hip fracture ward was set at five staff members for day shifts; on normal weekdays this consisted of two nurses and three assistants, whereas this was reduced to either one nurse and three assistants or two nurses and two assistants during weekends and holidays. Evening shifts were usually manned by one nurse and two assistants, but during weekends and holidays this was reduced to either one assistant and one nurse or two assistants and one nurse, with responsibility for both the hip fracture unit and another ward (0.5 nurse). Night shifts were manned by an assistant and a nurse with responsibility for two wards (0.5 nurse), regardless of the period. The unit had daily rounds by specially assigned orthopaedic trauma specialists on all weekdays, whereas rounds on weekends and holidays were usually performed by specialists or senior residents not necessarily attached to the hip fracture unit. Perioperative clinical problems were handled primarily by the junior orthopaedic resident. The unit had specially attached physiotherapists who worked with patients twice a day on weekdays, but only once daily during weekends and holidays. Chi-square test was used for testing the significance of categorical data and the Kruskal Wallis test for data in scales that was not normally distributed. After identification of significant risk factors, forward stepwise logistic regression was used to identify factors independently associated with early postoperative mortality. A factor was only entered in the model if it had P<0.10. The level of significance was set at P<0.05 with a power of All data analysis was performed with SPSS version 10.1 (Chicago, IL, USA). 451

3 Foss and Kehlet Results Of the 600 consecutively admitted patients 332 were admitted on a weekday, 118 during weekends and 150 during holiday periods. Five-day postoperative mortality was 1.8, 2.5 and 8.0% (P=0.02) and 30-day mortality was 11.1, 12.7 and 19.3% on weekdays, weekends and holidays, respectively (P=0.05). There were no significant differences with regard to patient age, cognitive or functional level or residential status, but patients admitted during weekend/holidays were significantly more often classified as ASA III/IV (Table 1). Patients admitted during weekends and holidays had more pertrochanteric fractures, but there were no significant differences in the delay to surgery, level of experience of the surgeon, length of surgery, lowest intraoperative mean arterial pressure or i.v. fluid infusions, intraoperative bleeding or number of perioperative transfusions. The risk factors for death within 5 and 30 days were analysed using uni- and multivariate analysis of all risk factors potentially associated with increased mortality from the initial analysis (Table 2). Age and functional level were independently associated with mortality at 5 and 30 days, whereas ASA classification was only independently associated with 30-day mortality, although a trend towards independent association was apparent for 5-day mortality. Dementia and nursing home residence was associated with mortality in the univariate analysis, but this association disappeared in the multivariate analysis. Delay to surgery for >24 h was univariately associated with early mortality, but was insignificant in the multivariate analysis. Admission during a holiday or vacation period was significantly and independently associated with death within 5 and 30 days of surgery with odds ratios of 4.34 and 1.84, respectively. Discussion Admission during a holiday period with reduced ward staffing levels was significantly and independently associated with a 4-fold increased risk of dying within 5 days of surgery and an almost doubled risk within 30 days. Mortality after weekend admission showed a non-significant increase compared with that of patients admitted on a weekday with normal staffing levels. Previous studies have not found any relationship between weekend admissions and mortality in hip fracture patients. 12 This has been attributed to the fact that hip fractures are relatively easy to diagnose and are perceived as requiring limited early perioperative resources, leading to the conclusion that they are not sensitive to the decreased staffing levels at weekends. In contrast, an increased mortality was demonstrated in patients with ruptured aortic aneurysm 1 and surgical patients admitted to an ICU during weekends. 3 Our study has several advantages. First, perioperative care principles were well defined regarding surgical Table 1 Perioperative characteristics of 600 consecutive hip fracture patients admitted to the hip fracture unit on a regular weekday, weekend or holiday. Values are presented as number of patients (percentage) for nominal data (x 2 -test) and as median (interquartiles) for data in scales (Kruskal Wallis test). NMS, new mobility score; MAP, mean arterial pressure; RBC, red blood cell; GA, general anaesthesia Day of admission Weekday Weekend Holiday N Age (yr) 82 (76 88) 83 (77 90) 83 (76 88) 0.26 Female 254 (77) 89 (76) 106 (71) 0.39 Poor functional level (NMS 0 5) 171 (52) 62 (53) 90 (60) 0.21 Dementia 90 (27) 33 (28) 48 (32) 0.54 Prefracture nursing home residence 62 (19) 23 (20) 36 (24) 0.39 In-hospital fracture 25 (8) 5 (4) 14 (9) 0.28 ASA classification (III IV) 180 (54) 75 (64) 98 (65) 0.04 Fracture type Medial 179 (54) 48 (41) 69 (46) Pertrochanteric 127 (38) 63 (53) 70 (47) 0.09 Subtrochanteric 24 (7) 7 (6) 9 (6) Pathological 2 (1) 2 (1) Delay to surgery (h) 19 (15 23) 18 (15 23) 18 (13 22) 0.22 Experience level of surgeon Orthopaedic specialist 106 (32) 52 (44) 57 (38) Senior registrar 57 (17) 15 (13) 21 (14) 0.18 Junior registrar 169 (51) 51 (43) 72 (48) Duration of surgery (min) 53 (40 70) 59 (40 80) 49 (30 64) 0.01 Epidural analgesia 321 (97) 113 (96) 144 (96) 0.87 GA (supplementary or alone) 32 (10) 10 (9) 13 (9) 0.53 Intraoperative MAP <60 (mm Hg) 98 (30) 30 (30) 37 (25) 0.52 Intraoperative bleeding (ml) 200 ( ) 200 ( ) 200 ( ) 0.08 Intraoperative i.v. fluids (ml) 1000 ( ) 1000 ( ) 1000 ( ) 0.25 Total RBC transfusions (units) 2 (0 3) 2 (0 4) 2 (0 3) day mortality (after operation) 6 (2) 3 (3) 12 (8) day mortality (after operation) 37 (11) 15 (13) 29 (19) 0.05 In-hospital mortality 30 (9) 12 (10) 25 (17)

4 Hip fracture mortality after holiday admission Table 2 Relationship between postoperative 5- and 30-day mortality and perioperative patient characteristics in 600 consecutive elderly hip fracture patients treated according to a multimodal rehabilitation regimen. NMS, new mobility score; NA, not applicable Death within 5 days after operation Death within 30 days after operation Univariate analysis Multivariate analysis Univariate analysis Multivariate analysis >85 yr 5.42 ( ) < ( ) ( ) < ( ) <0.001 Female sex 0.84 ( ) 0.71 NA 0.96 ( ) 0.87 NA Prefracture NMS ( ) ( ) ( ) < ( ) 0.01 Dementia 2.88 ( ) ( ) < Nursing home residence 1.83 ( ) ( ) ASA score III/IV 4.37 ( ) ( ) < ( ) <0.001 Pertrochanteric fracture 1.46 ( ) 0.39 NA 1.18 ( ) 0.48 NA >24 h delay to surgery 2.93 ( ) ( ) 0.46 NA Holiday admission 4.26 ( ) ( ) ( ) ( ) 0.02 delay, anaesthesia, analgesia, nutrition, fluid, transfusion and physical therapy. 6 Second, staffing levels were defined in detail and, third, we were able to code the admission of the patients specifically for days that had planned reductions of staff because of weekends and holidays. The standardized care principles were implemented during all days of the year, regardless of weekends and holidays. There were no planned or recorded discrepancies in experience of the attending staff members, or the standards of given care between weekdays and weekends/holiday periods when pre- and intraoperative care levels were examined. Patients were admitted, anaesthetized and operated on by a comparative mix of experienced staff members, regardless of the time of admission, and delay to surgery, operating time, bleeding, fluid therapy, number of transfusions and intraoperative blood pressure did not show significant differences between the groups. However, the postoperative level of care was, potentially, more affected by both weekends and holidays. During these periods, the number of staff members available per patient in the ward was reduced by 20% on day shifts, and the number of nurses available during day and evening shifts was reduced by between 33 and 50%. The amount of physiotherapy staffing was reduced by 50% on weekends and holidays. Finally, postoperative rounds were not performed by the assigned orthopaedic trauma specialists, which potentially decreased the standard of care in these complex patients. The impact of reduced staffing in the postoperative setting is potentially huge, despite a standardized care setup. The staff still applied standards of care (see Patients and methods section) such as the prescribed anaesthesia and analgesia, antithrombotic prophylaxis, transfusions and nutritional supplementation. However, the quality of the practical application of care standards may have been reduced. Thus, standards may have been reduced. Thus, patients may have had unrecognized complications, or a delay in the diagnosis of these, because of reduced nursing availability. Hypovolaemia and dehydration could, potentially, go unnoticed for a longer time and the degree of mobilization may be reduced. Although all these factors are hard to quantify compared with the more easily recognized preand intraoperative factors (Table 1), their potential impact on postoperative outcome may support our conclusions. Physiotherapy was not measured directly, but it would seem reasonable to assume that the 50% reduction in physiotherapists during weekends/holidays would lead to a reduction in the amount delivered. No previous studies have looked at mortality in acute surgical patients with admission during longer periods of decreased hospital staffing such as during winter and summer vacation periods. A July effect has been hypothesized in American studies based on influx of junior doctors rather than staff reductions, but this did not influence outcome in trauma patients. 8 Seasonal variation in outcome after hip fracture has been reported previously, with mortality peaking during winter months. 9 The present study incorporates a mixture of holiday periods during both summer and winter months and, as such, the present findings cannot be explained by a seasonal variation. Mortality at both 5 and 30 days was significantly higher during holidays periods compared with weekends and weekdays. This may be interpreted as a consequence of decreased level of perioperative nursing care, including physiotherapy, rather than a simple delay in diagnosis and treatment as has been suggested in other diseases, where a weekend effect on mortality has been found. 1 This is also consistent with previous studies, where an increased mortality was demonstrated in surgical patients when the nurse:patient ratio decreased. 10 Our definition of weekends was based on the availability of the trauma theatre in our unit. We defined weekends as beginning Friday afternoon, as this would include patients receiving their operation on Saturday and Sunday in the weekend group, while patients admitted after Sunday afternoon were operated on Mondays. Previous studies have defined weekends, more or less arbitrarily, as starting either on midnight Friday or noon Saturday. 12 A possible reason for our study not having a weekend effect could be that 453

5 Foss and Kehlet >48 h of decreased staffing levels in the ward are required to have a negative effect on outcome. We hypothesized that the effect of weekend/holiday admissions would be best discernible from other factors contributing to postoperative mortality during the first few days after surgery, as demonstrated in other conditions. 1 Our findings are similar to studies on the effect of regional anaesthesia, 11 where the positive effect on early mortality was blurred at 30 days and later by the heterogeneity, age and general fragile condition of the hip fracture population. In conclusion, our study demonstrates that patients with a hip fracture had an independently increased risk of early postoperative mortality when admitted during longer holiday periods where hospital/ward staffing was reduced. The role of staff experience and availability need detailed assessment in future studies. Acknowledgements This work received financial support from IMK Fonden (Copenhagen, Denmark). N.B.F. hypothesized, data gathered, data analysed, prepared the manuscript, approved the final draft and H.K. hypothesized, critically reviewed and approved the final draft. References 1 Bell CM, Redelmeier DA. Mortality among patients admitted to hospitals on weekends as compared with weekdays. N Engl J Med 2001; 345: Cram P, Hillis SL, Barnett M, Rosenthal GE. Effects of weekend admission and hospital teaching status on in-hospital mortality. Am J Med 2004; 117: Ensminger SA, Morales IJ, Peters SG, et al. The hospital mortality of patients admitted to the ICU on weekends. Chest 2004; 126: Kehlet H, Dahl JB. Anaesthesia, surgery and challenges in postoperative recovery. Lancet 2003; 362: Foss NB, Kristensen MT, Kristensen BB, Jensen PS, Kehlet H. Effect of postoperative epidural analgesia on rehabilitation and pain after hip fracture surgery: a randomized, doubleblind, placebo-controlled trial. Anesthesiology 2005; 102: Foss NB, Kehlet H. Mortality analysis in hip fracture patients: implications for the design of future outcome trials. Br J Anaesth 2005; 94: Parker MJ, Palmer CR. A new mobility score for predicting mortality after hip fracture. J Bone Joint Surg Br 1993; 75: Busse JW, Bhandari M, Devereaux PJ. The impact of time of admission on major complications and mortality in patients undergoing emergency trauma surgery. Acta Orthop Scand 2004; 75: Crawford JR, Parker MJ. Seasonal variation of proximal femoral fractures in the United Kingdom. Injury 2003; 34: Aiken LH, Clarke SP, Sloane DM, Sochalski J, Silber JH. Hospital nurse staffing and patient mortality, nurse burnout, and job dissatisfaction. JAMA 2002; 288: Valentin N, Lomholt B, Jensen JS, Hejgaard N, Kreiner S. Spinal or general anaesthesia for surgery of the fractured hip. Br J Anaesth 1986; 58: