Medical and economic consequences of perioperative complications in older hip fracture patients

The present study is a sub-analysis of a prospective observational study including a total of 402 patients with hip fractures treated between April 2009 and October 2011. Patients ≥ 60 years with operative treatment due to hip fractures were included. Exclusion criteria were malignancy-related fractures and polytrauma patients (ISS ≥ 16 [19]). Five hundred thirty-nine patients with fracture were treated at our hospital during the observation period. Four hundred seventy-seven patients met the inclusion criteria. Of those 477 patients, 75 patients declined participation. We obtained approval by the local Ethics committee (AZ 175/08). Each patient or their legal guardians provided written informed consent for participation in the study. Mortality, clinical course, treatment costs, and short-term follow-up are already published in some publications of our study group [13, 20‐22].

The following baseline data were documented from all patients: age, fracture classification (nondisplaced femoral neck fractures, displaced femoral neck fractures, stable pertrochanteric fractures, unstable pertrochanteric fractures, and subtrochanteric fractures), American Society of Anaesthesiologists (ASA) classification [23], Charlson Comorbidity Index [24], and Mini-Mental State Examination [25].

The following data were collected during time of hospitalization: time to surgery (≤ 24 h vs. > 24 h), surgical procedure (screw fixation, intramedullary nail, dynamic hip screw, hemiarthroplasty, and total hip arthroplasty), ICU readmission after surgery, transfusion of erythrocyte concentrates, and duration of stay at the acute-care hospital. Follow-up was obtained after 6 months and 1 year. EQ5D-5L [26], Barthel score [27], and Tinetti score [28] were collected during the different follow-up examinations. The Barthel Score (range 0–100 points) records essential everyday functions of the patients. Patients are divided into independent, partially independent, needy and largely dependent on care. The Tinetti score (range 0–28 points) is divided into 2 parts, a balance test and a walking test. The Tinetti score can be used to quantify the fall risk of the patient. Both tests are established screening tools in the treatment of geriatric patients. Complications were assessed during the inpatient stay and classified by Clavien and Dindo. The current classification is an update of the previous classification by Clavien in 1992. It was published in 2004 by Dindo et al. Grade 1 complication represents any deviation from the normal without the need of pharmacological treatment (except for antiemetics, antipyretics, analgetics, diuretics, electrolytes, and physiotherapy) or surgical, endoscopic, and radiological interventions. This includes for example transient elevation of the serum creatinine, atelectasis requiring physiotherapy, and noninfectious diarrhea. Grade 2 complication stands for a pharmacological treatment with drugs other than such allowed in grade 1, for example tachyarrhythmia requiring β-receptor antagonist, pneumonia, and urinary tract infections treated with antibiotics. Grade 3 complications require surgical, endoscopic, or radiological intervention like wound infection with the need of surgical intervention with the need of an anesthesia or pleural effusion with the need of a pleura drainage. Life-threatening complications are defined as grade 4 complications This includes for example acute renal failure requiring dialysis lung failure requiring intubation [29].

The methodology regarding the calculation of the costs that has already been described in a prior study of our group [30]. All potential cost factors were recorded as accurately as possible for each of the 402 patients individually. Afterwards, the actual costs for each cost factor were retrospectively calculated with help of the controlling division in our financial department. For each patient, up to 196 individual parameters were recorded [30].

Data were collected in a FileMaker® database (FileMaker Inc., Santa Clara, CA, USA). We performed double entry with a plausibility check to monitor data quality. IBM SPSS Statistics 24 (Statistical Package for the Social Sciences, IBM Corporation, Armonk, NY, USA) was used for statistical analysis. To identify possible influencing factors, bivariate analysis was conducted using Student’s t test or ANOVA. Statistical significance was assumed at p < 0.05. Finally, multivariate analysis with backward selection was performed to detect independent influencing factors (inclusion at p < 0.05, exclusion at p > 0.10).

The mortality of patients without complications was 13.7% in the 6 months’ follow-up and 22.3% in the 12 months’ follow-up. Patients suffering from type III complications had a significant higher mortality (27.7%; p = 0.018) after 6 months compared to patients not suffering from any of these complications. In the 12 months’ follow-up, this trend could be detected as well (35.9%), however without statistical significance (p = 0.072). Complications type II and IV according to Clavien and Dindo showed no significant influence on mortality after 6 or 12 months (Table 2).

Type II (0.56 (min. − 0.2; max. 1.0); p = 0.032) as well as type III (0.46 (min. 0.0; max. 1.0); p = 0.010) and type IV (0.41 (min. 0.0; max. 0.9); p = 0.005) complications were associated with a significant decrease of EQ-5D after 6 months compared to no complications (0.65 (min. − 0.2; max. 1.0)). However, after 12 months, only patients suffering from type IV complications (0.42 (min. 0.1; max. 0.9); p = 0.017) had a significantly lower EQ-5D compared to patients not suffering from any complication (0.66 (min. − 0.2; max. 1.0). A trend towards a reduced EQ-5D after 12 months was detected in patients suffering from type II complications; however, this was not statistically significant (p = 0.051). No impact of type III complications was found after 12 months (Table 3).

Patients not suffering from a complication had a Barthel score of 75.4 (min. 0; max. 100) after 6 months and 75.6 (min. 0; max. 100) after 12 months. Patients suffering from type II complications during hospitalization had a significantly lower Barthel score after 6 months (61.3 (min. 0; max. 100); p = 0.001) and 12 months (58.7 (min. 0; max. 100); p = 0.001). Similar results were found in patients who suffered from type IV complications after 6 months (41.7 (min. 5; max. 100); p < 0.001) and 12 months (50.8 (min. 5; max. 100); p = 0.013). Patients suffering from type III complications showed a trend to a lower Barthel score after 6 months (62.7 (min. 10; max. 100); p = 0.075), while after 12 months the Barthel score increased to (72.6 (min. 15; max. 100); p 0.770) (Table 4).

Complications types II–IV (type II, 14.1 (min. 1; max. 28); p = 0.006; type III, 13.3 (min. 1; max. 28); p = 0.043; type IV, 9.3 (min. 0; max. 28); p = 0.002) showed a significant reduction in the Tinetti score after 6 months compared to no complication (17.4 (min. 1; max. 28)). Twelve months after the fracture, type II (14.0 (min. 1; max. 28); p = 0.002) and type IV complications (8.9 (min. 1; max. 23); p = 0.002) were associated with a significantly lower Tinetti score compared to patients not suffering from any of these complications (18.3 (min. 0; max. 28)). No significant influence was detected in patients with type III complications (Table 5).

After adjusting for age, gender, and ASA score, a trend towards an increased mortality after 6 months was seen after type III complications (p = 0.066), whereas other types of complications did not show associations. HRQL was significantly reduced after type II–IV complications at 6 months’ follow-up (p values: type II, 0.046; type III, 0.010; type IV, 0.010). After 12 months, a trend towards reduced HRQL was seen after type II (p = 0.069), no association after type III and a significantly reduced HRQL after type IV complications (0.019). Type II and IV complications were significantly associated with a reduced Barthel score and Tinetti score after 6 and 12 month (p values: Barthel score: type II, 6 months, 0.001; 12 month, 0.001; type IV 6 months, < 0.001; 12 months, 0.010; Tinetti: type II 6 months, 0.005; 12 months, 0.010; type IV 6 months, 0.001; 12 months, 0.002). A trend towards reduced Barthel and Tinetti scores was seen after type III complications after 6 months (p values: BI, 0.058; Tinetti: p = 0.070), while after 12 months no more associations were observed.

The average costs for overall acute-care treatment were 8853€. All types of complications resulted in significantly higher treatment costs. Acute care costs of patients suffering from no complications averaged 7814€ (95%CI 7291–8338 €). A significant increase was noted in patients suffering from type II complication (9380€; (95%CI 8192–10,569 €) p = 0.004), type III complications (12,782€; (95%CI 10,014–15,550 €) p < 0.001), or type IV complications (13,635€; (95%CI 9400–17,870 €) p < 0.001) (Fig. 1).

Several authors investigated mortality after hip fractures in older patients. Most of these studies focused on patients’ characteristics and data from hospitalization. Usually postoperative complications were not investigated. ASA-Score [31, 32], comorbidities [31, 33], duration of postoperative immobilization [31], gender [32‐34], age [32, 33], and surgical delay [32] have been shown to significantly increase mortality after hip fracture. In this study, predominantly type III complications increased mortality. This underlines that mainly those complications requiring surgical revision seem to increase the risk of dying in a 12-month follow-up after surgical fixation of hip fractures.

Although several studies investigated the influence of hip fractures on quality of life, to the best of our knowledge few have included the influence of complications. Alexiou et al. showed in a review that the majority of those patients did not regain their pre-fracture quality of life [35]. Reviewing 49 studies, they stated comorbidities, female gender, and undernutrition as factors associated with a negative impact on health-related quality of life. These factors also lead to longer duration of hospital stay, severe postoperative pain, and complications [35]. Taking a closer look at the occurrence of complications, all types resulted in a significantly lower EQ-5D 6 months after the fracture. No statistically significant difference of HRQL was found between patients suffering from type III complications and those not suffering from any complication at 12 months’ follow-up. In contrast, HRQL in patients with type II or IV complications does not equalize during the 12 months follow up. It appears that patients suffering from type III complications experience notable reduction of HRQL in the short term; however, they show a high potential of recovery. In contrast, no further improvement between 6 and 12 months was seen in patients who have suffered from a type II or IV complication. This suggests that type II and IV complications tend to reflect a worsening of the general condition and an increase in secondary diseases, while typical type III complications, such as surgical revision, do not necessarily reflect the patient’s state of health. Hansson et al. showed in their retrospective cohort study similar results concerning patients suffering from complications. They found that the absence of complications was associated with higher satisfaction and lower pain, even 6 months after surgery [36].

Patients suffering from type II and type IV complications had a significantly lower Barthel score 6 and 12 months after surgery. Patients suffering from type III complications showed a trend to a lower Barthel score 6 months after surgery, which was not significant. However, they almost regained their functional capabilities after 12 months compared to patients without any complications. The additional surgery seems to prolong functional recovery of those patients, but they are not significantly affected at the 1-year follow-up. In contrast, type II and type IV complications may represent a surrogate parameter for the frailty of the patients like mentioned already above. Several authors investigated functional outcome after hip fracture. Most of them focused on patients’ characteristics and data from hospitalization but did not examine the influence of different types of complications on the mid-term outcome. In the recent literature, inconsistent results have been published by different authors. This might be due to the different study collectives and study protocols. Patient characteristics that have been associated with poor functional outcome by many authors are increased age [37‐40], poor mental status [37, 39, 41, 42], and higher degree of comorbidity [38]. While there are now contradictory statements about high age alone as a risk factor, there is agreement about the influence of the previous state of health [43].

“Tinetti score” was used to illustrate the influence of complications on mobility after hip fracture. All types of complications lead to a significantly lower Tinetti score compared to patients not suffering from complications 6 months after the surgery. Similar to HRQL and functional status patients with type III complications improved after 12 months, whereas type II and IV complications were associated with a significantly reduced Tinetti score also after 12 months. Again, a possible explanation could be that patients suffering from surgical complications require a couple more months for rehabilitation. Nevertheless, once survived patients suffering from type III complications achieve comparable mobility to patients without complications whereas mobility in patients with type II and IV complications remains significantly decreased. Again, these complications seem to be a surrogate parameter for an overall poor general health status, which could be reasonable for reduced mobility.

Our group already showed that ASA-Score, Charlson Comorbidity Index, and fracture location are influencing factors that lead to increased costs of acute care for hip fracture treatment [30]. Complications during in-hospital treatment do not only have an impact on patient’s mobility, mortality, functional capability, and health-related quality of life; they go along with significantly increased acute care costs as well. Broderick et al. showed in their selective review regarding failure of geriatric fractures that implant failure does not only lead to a twofold increase in length of stay, but leads to a doubling of healthcare costs as well [44]. Other authors showed increased costs after implant failure, wound infection, others reasons of reoperations, and medical complications [17, 18, 44, 45]. In line with these findings, type III complications were associated with increased costs in the present study. However, patients who needed surgical revision were not those causing the highest costs in our study sample. Patients suffering from type IV complications were associated with the highest healthcare costs. This might be due to prolonged intensive care treatment and the necessary intensive inpatient treatment after suffering from these complications (e.g., dialysis). As expected, type III complications increase healthcare costs in a significant amount. Even type II complications increase healthcare costs during inpatient treatment significantly. This is explained by an extra amount of medical or personal capacities (e.g., drugs like antibiotics) required by patients suffering from type II complications.

The results of the present study are limited by some factors. We included only patients from one single center, and only selected parameters were collected and analyzed. The fact that orthogeriatric treatment, which has been proven to be effective in reducing complications, was not performed in the present analysis represents another limitation. Furthermore, a cost analysis, regardless of how detailed and accurate it is, contains some assumptions. For example, it is not possible to exactly determine the duration of physician or nurse-patient contact during the acute ward phase. Another drawback of this study is that it only includes primary hospitalization costs. Finally, the age restriction ≥ 60 years has to be mentioned as a limitation, as this is not the typical definition of a geriatric patient and therefore non-geriatric patients may have been included.

Nevertheless, the prospective design and its large patient collective are strengths of this study. Furthermore, the cost analysis represents the result of an extremely detailed analysis of the actual costs of care for every individual patient.