This is a retrospective case–control study to ascertain the factors influencing nosocomial infection (NI) in elderly patients with hip fractures.
Methods
A total of 80,174 patients (≥ 60 years) who suffered hip fractures between 2006 and 2017 were identified through a national inquiry of 94 hospitals. The patients were divided into an NI group and control group according to the occurrence or lack of occurrence of NI within 48 hours after surgery, respectively. Age, gender, hip fracture pattern, whether to operate, surgical treatments, and comorbidities were recorded as variables.
Results
A total of 9806 elderly hip fracture patients (60 years) were included, 1977 of whom were patients diagnosed with NI. The control group consisted of randomly drawn cases from the 9806 patients from different hospitals with a rate of one NI patient: four patients without NI. Patient gender, age, and in particular the number of comorbidities were associated with occurrence of NI. Using regression models to predict infection outcomes based on the number of comorbidities had an area under the curve (AUC) of 0.714, while using the Charlson comorbidity index (CCI) yielded a smaller value of 0.694. The most common comorbidities of this elderly cohort were chronic respiratory disease, hypertension, diabetes mellitus, cerebrovascular disease, and coronary heart disease.
Conclusions
Older age, male gender, and greater number of comorbidities were found to be associated with the occurrence of NI. In particular, the number of comorbidities was the most accurate predictor of NI occurrence, and when used to build a regression model, it had greater predictive capability than CCI to predict NI in elderly hip fracture patients. Additionally, the common diseases of the elderly should be primarily considered when investigating the relationship between comorbidities and NI in older patients.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Introduction
Hip fracture is one of the most common fractures in older people, and hip fractures in the elderly often lead to adverse post-operative events and poor prognosis [1‐5]. Furthermore, with the overall aging of the human population, the number of elderly hip fracture patients in Asia will account for 50% of the world by 2050 [6]. Nosocomial infection (NI) is the most common post-operative adverse event in elderly patients with hip fractures [7]. Advanced age, comorbidities, and other clinical factors contribute to the occurrence of NI in the elderly, which subsequently results in prolonged hospital stay, increased cost, and poor prognosis [2, 5, 8‐10]. Naturally, clarifying the association between clinical variables and occurrence of NI is imperative to improving the prognosis of elderly patients with hip fracture.
Previous studies on NIs in elderly patients who receive surgical treatment of hip fractures have mainly focused on surgical site infections (SSI) [11]. Factors such as age [10], comorbidities[12, 13], surgical methods [14], season [15], and delayed surgery [12, 16, 17] have been found to be related to the occurrence of SSIs. Because of their catastrophic consequences, SSIs have drawn a large amount of attention from clinical surgeons. However, other infections, such as lung infections[18], urinary tract infections[7, 19], and skin pressure ulcers[20, 21], account for the majority of NIs in elderly patients after surgery. Moreover, the occurrence of NIs in organs outside of surgical sites leads to a decline in the quality of life of patients as well as increases in hospital readmission and mortality. Cruz et al. reported that age as an independent risk factor was positively related to the incidence of NIs in elderly patients who receive surgical treatment for hip fractures [10]. Another study found that elderly hip fracture patients with multiple comorbidities tended to require longer hospital stays which thus led to a greater probability of NI occurrence [22].
Anzeige
This study retrospectively analyzed the inpatient data records of elderly hip fracture patients from 194 Chinese hospitals nationwide, aiming to elucidate the risk factors of NI in elderly hip fracture patients. We hypothesized that age, comorbidities, and other clinical factors might contribute to the occurrence of NI in elderly hip fracture patients. Furthermore, we investigated both the severity and types of comorbidities in relation to the occurrence of NI, which we will refer to as “comorbid condition,” with the aim to provide suggestions to help prevent occurrence of NI in elderly hip fracture patients.
Materials and methods
Medical record data from the homepage of inpatients of elderly inpatients with hip fractures was collected from 194 Chinese hospitals nationwide between 2006 and 2017. All data were collected and audited by trained researchers according to the standards of the ICH GCP and ICH E9 [23, 24].
Age, gender, fracture pattern, whether to operate, surgical treatments, and comorbidities were recorded as variables. All patients older than 60 with hip fractures were enrolled and subdivided into the age ranges, 60–69, 70–79, 80–89, and 90 + . The hip fracture patterns in this study included femur neck and intertrochanteric, regardless of the specific clinical type. We classified surgical treatment as arthroplasty (either a partial or a total hip replacement) and osteosynthesis. All patients with hip fractures were treated following formal clinical principles and practices. Comorbidities were diagnosed according to the admission information provided in the data. The number of comorbidities was used to briefly describe the condition of comorbidity of individual patients [13, 25]. Using “the number of comorbidities” as a proxy for health of patients was simpler than including specific and complex disease categories in our statistical analyses of clinical scenarios. Considering that the Charlson comorbidity index (CCI) [26] has been widely applied to measure the severity of comorbidities, two regression models using (1) the number of comorbidities or (2) CCI were established separately to compare which was more effective in predicting the occurrence of NI. The specific types of comorbidities recorded in this study included chronic respiratory disease, hypertension, cerebrovascular disease, diabetes mellitus, coronary heart disease, cardiac failure, anaemia, dementia, gastrointestinal haemorrhage, peripheral vascular disease, peptic ulcer, renal disease (mild and moderate–severe), liver disease (mild and moderate–severe), and malignant tumor. All of the comorbidities were classified according to the principles of the CCI [26]. Nosocomial infection, the outcome in these models, was defined as any infections (either at the surgical site, pneumonia, urologic infection, or others) that were acquired during hospitalization after surgery. In this study, the diagnosis of NI was based on a retrospective analysis of medical records, and patients with any clear diagnosis of infections at the time admission or within 48 hours after admission were excluded.
In this case–control study, patients were grouped based on whether or not they were diagnosed with NI. All patients with NI were collected from the hip fracture patient cohort as the observation group. Patients without NI were stratified and randomly selected from the cohort as the control group. To control for the differences between hospitals from different regions, the hospitals where the NI case was from were used as match factors. The match ratio of the control group was 1 NI patient: 4 patients without NI from a given hospital to improve statistical accuracy as much as possible.
Anzeige
Statistical analysis
We have used subgroups of age ranges to treat age as a categorical variable instead of a continuous variable. All categorical variables were expressed as percentages and compared using the chi-squared test. Pearson chi-square was used when the expected frequency was ≥ 5 and the total sample size was ≥ 40. Continuous calibration chi-square test was used when expected frequency was < 5 but ≥ 1, and total sample size was ≥ 40. Binary logistic regression models were used to examine the relationships of all variables including age range, gender, fracture pattern, whether to operate, surgical treatment, the number of comorbidities, and CCI with occurrence of NI. The results are presented in crude and adjusted odds rate (OR), 95% confidence intervals (CI), and p-values. A univariate analysis was performed using a simple logistic regression model to estimate the crude OR and the corresponding 95% CI. Multiple logistic regression analysis was then performed to adjust the effects of each variable. Considering that the surgical treatment was dependent on the type of fracture pattern, the logistic regression of surgical treatment was performed after stratification of patients based on hip fracture patterns. The significance level for all statistical tests was set at 5% (P < 0.05). Statistical analyses were performed using SPSS 20.0 statistics software (IBM Corporation, Chicago, IL).
Results
A total of 9806 elderly hip fracture patients were included in this study. The NI group was composed of 1977 patients with NI from the 80,174 elderly hip fracture cohort. Another 7829 elderly hip fracture patients without NI were also randomly selected from the same cohort accordingly as a control group (case–control = 1:4).
Comparison between groups of single variables
The frequency distributions of all variables in the NI and control groups and the frequency of suffering from NI in relation to each range within the different variables were listed in Table 1. A higher proportion of males were found in the NI group, and the frequency of NI in males was significantly higher than that in females (24.2% vs.17.6%, P < 0.05). Additionally, the overall proportion of patients in the NI group compared to the control group increased with age. For example, in 80–89, 41.2% of the patients were in the NI group, while 30.2% were in the control, and in 90 + , 7.8% of the patients were in the NI group, while 5.2% were in the control. The frequency of NI increased as the age ranges increased. The distribution of the types of NIs in each age range were shown in Supplementary Table 1 (Online Resource 1). Femur neck fracture was more common than intertrochanteric fracture in both the NI and control groups, and the frequency of suffering from NI in patients with femur neck fracture was lower than that of the patients with intertrochanteric fracture (19.0% vs. 21.6%, P < 0.05). Similarly, fewer patients underwent operation for treatment than those that did not undergo operation in both groups, but the frequency of NI in the two groups was opposite (24.1% vs. 17.9%, P < 0.05). In regard to the surgical treatment variable, there was a significant difference in the frequency distribution of arthroplasty and osteosynthesis, but no significant difference was observed in the frequency of suffering from NI between the two types of surgical treatments (P > 0.05). In the NI group, except for 0 comorbidities, the proportion of patients suffering from 1 to 3 comorbidities was significantly higher than that in the control group and the frequencies of suffering from NI increased gradually as the number of comorbidities increased (P < 0.05). Likewise, similar results were also observed using CCI (Online Resource 2). Additionally, the in-hospital death rate of the NI group was significantly higher than that in the control (P < 0.05). From the comparison between the NI and control group with respect for each type of comorbidity, almost all kinds of comorbidities recorded here had significant differences in frequencies, except for moderate–severe renal disease, peripheral vascular disease, peptic ulcer, and moderate–severe liver disease. Moreover, in the patients with each kind of comorbidity, the frequencies of NI were higher overall than the patients without the comorbidity (P < 0.05, except for the aforementioned comorbidities.)
Table 1
The comparison of all variables between nosocomial infection group and control
The results of the multivariate analysis were shown in Table 2. In the simple logistic regression analysis of the whole sample, all variables were associated with NI. In particular, the logistic regression of surgical treatment was performed with the operation patients who were stratified according to different fracture patterns. After adjusting the multiple logistic regression, gender, age range, and the number of comorbidities were considered associated with NI. Fracture pattern, whether to operate, and surgical treatment had no significant association with NI. The adjusted ORs and 95% CIs were shown in Fig. 1. Furthermore, the regression model above was tested and showed an AUC value of 0.714, and the Hosmer–Lemeshow (H–L) test p-value was 0.783. Using the CCI instead of multiple logistic regression with the number of comorbidities yielded similar results (Online Resource 3). However, according to testing of the CCI, results showed an AUC of 0.694 and the H–L test P = 0.896.
Table 2
Simple and multiple logistic regression showing the effect of different variables on nosocomial infection
Adjusted OR of multivariable binary logistic regression (gender, age, and the number of comorbidities were significantly associated with nosocomial infection)
×
Comorbid conditions
Comorbid conditions were analyzed to explore their specific influences on NI occurrence. Figure 2 showed the frequency distributions of patients with different numbers of comorbidities in different stratifications of other variables. The comparison of comorbidity frequencies between different surgical treatments had no significant differences (P > 0.05), so the results were not included in Fig. 2. As age ranges increased, not only did the total frequency of comorbidities increase (P < 0.05, including all patients with 1–3 comorbidities), but also the proportions of patients with each comorbidity also increased. In regard to gender, male patients constituted a significantly higher proportion of patients at each level of comorbidity severities than female patients (P < 0.05). In 0–2 comorbidity levels, patients who suffered from femur neck fracture showed a significantly higher frequency of comorbidities than those who suffered from intertrochanteric fracture (P < 0.05). Additionally, the proportion of individuals without comorbidities in patients who underwent operations were significantly higher than the patients receiving conservative treatments (P < 0.05). In order to further clarify the impact of the number of comorbidities on NI occurrence, the composition of comorbidities in different comorbidity severities was analyzed. The top five most frequent comorbidities or comorbidity groups, were shown in Fig. 3, annotated with their frequencies.
Fig. 2
The relationship between the number of comorbidities and the other variables. The stacked bars are labeled with the frequency distribution of patients with different numbers of comorbidities in different stratifications
Fig. 3
The comorbid conditions of patients with different numbers of comorbidities. Abbreviations: CRD, chronic respiratory disease; HT, hypertension; DM, diabetes mellitus; CVD cerebrovascular disease; CHD, coronary vascular disease
×
×
Discussion
NI has been considered as the most common adverse post-operative event in orthopaedics [7, 22, 27]. For the elderly with hip fracture, because of their advanced age and complex comorbidities, NI consists of infections not only at surgical sites (SSI) but also in other organs, such as pulmonary infections, urinary tract infections, and skin pressure sore infections. These infections have been proven to be associated with the increased mortality and higher hospital costs [7, 18‐21, 28‐30]. Age, gender, surgical factors, and comorbidities have been investigated in association with SSIs [14, 15, 28, 31]. However, to date, the factors associated with the occurrence of NI, which includes SSI and infections in other organs, have remained unclear [10]. This study sought to clarify the factors associated with the occurrence of NI in elders with hip fracture using a multivariate regression analysis of data obtained from 194 hospitals across China, which is a larger and more generalizable patient population than previous studies [10, 19, 20, 29].
In this study, gender, age range, fracture pattern, whether to operate, surgical treatment, number of comorbidities, in-hospital death, and CCI all showed significant differences between the NI and control groups. In particular, patients with NI had a significantly higher in-hospital mortality than those without NI, which was consistent with previous studies [1, 4, 18]. This result also supported our viewpoint that considering NI as a vital and critical event is essential for supporting the elderly with hip fracture. Additionally, the frequency of NI in patients who died in hospital was almost 3.7 times higher than patients who survived (67.1% vs. 18.3%).
Multiple logistic regression results showed that gender, age range, and the number of comorbidities were associated with NI occurrence which was consistent with our hypothesis. However, fracture pattern, whether to operate, and surgical treatment were not significantly associated with NI occurrence. In a further analysis using a stepwise logistic regression model, fracture pattern was excluded when the age range was included as an adjustment, and whether to operate was excluded when the number of comorbidities was included. In fact, the fracture pattern type has been found to be associated with age in previous research [10, 32]. The decision of whether or not to operate might also be representative of the comorbid conditions of elderly patients; elderly patients with more comorbidities might not choose to undergo surgery [33]. Therefore, the lack of significant results about whether or not to operate was understandable when considering the actual situation. The female gender was found to be a protective factor in this population, which was different from previous studies of NI in the elderly with hip fracture [10, 14, 15]. However, some studies which focus on post-operative SSI and pneumonia have shown similar results [18]. Thus, these contradictory results might not be caused by gender itself but by other gender-related factors. The specific reasons may need to be clarified further through detailed targeted hierarchical modelling and research. In a previous study, age has been shown to be a predictor in NI occurrence in older patients with hip fractures who underwent operations [10]. Our study was not only consistent with these previous results, but also a gradual increase in frequency of NI occurrence as age range increased was also observed. Moreover, the number of comorbidities was found to be the strongest risk factor in this study. Previously, Cruz et al. reported that the severity of comorbidities measured by CCI showed no significant association with NI [10]. Considering the larger sample size used in our analysis and the relatively uniform distribution of patients with different comorbidity severities, it is reasonable that we have found different results from the aforementioned previous study. Additionally, the occurrence of SSI has also been previously shown to be associated with comorbidities, which provides indirect evidence in support of our results.[14, 15, 28].
Anzeige
The results of the regression models tests showed that the factor of the number of comorbidities had a stronger prediction capability for the occurrence of NI than CCI. To clarify this, the relationship between the number of comorbidities and CCI was analyzed in Fig. 4. According to the map, the CCI score did not correspond to the number of comorbidities, and the patients with the same number of comorbidities had different CCI scores. CCI is originally intended to predict mortality through the condition of comorbidities, so the types of comorbidities in the CCI analysis tended to significantly associate with mortality rather than with NI [26, 34]. This could be supported by the results of comparison of comorbidities between the NI and control groups (Table 1). Some comorbidities not included in CCI, such as hypertension and anemia, were associated with NI, while other comorbidities in CCI, such as peptic ulcer, peripheral vascular disease, moderate–severe liver diseases, and moderate–severe renal diseases, were not.
Fig. 4
Heat map of the relationship between the number of comorbidities and CCI
×
In further analysis of the number of comorbidities, we found that with higher age range, the number of comorbidities the patients would have was greater. Consequently, when analyzing the results of logistic regression, we treated age as an adjusted variable, and the number of comorbidities was used to predict the occurrence of NI. The frequency distribution of patients with different numbers of comorbidities in gender stratification indicated that the comorbid conditions of elderly male patients were more serious than those of females, which might explain the significant association between the gender and NI variables in the logistic regression model. Although the type of fracture pattern had no significant associations with the occurrence of NI, the frequency distribution of patients with different numbers of comorbidities had significant differences between femur neck fracture patients and intertrochanteric fracture patients. This finding might provide a basis for future research on the causes of different types of fractures. In addition, the results regarding whether or not to operate reflected that conservative treatments were more likely to be selected by surgeons for elderly patients with severe comorbid conditions. The specific enumeration of comorbid conditions indicated that the most common chronic diseases consisted of a major part of the comorbidities in elderly patients with hip fracture. Additionally, chronic respiratory disease, diabetes mellitus, hypertension, and cerebrovascular disease, which were the top five most frequent, have been found to be associated with post-operative infection in both the surgical site and other organs [14, 18, 20, 27, 35, 36]. Moreover, some diseases which were not in the top five most frequent, such as anemia, malignant tumour, and cardiac failure, have also have been found to be associated with post-operative infections [35, 37, 38]. Therefore, when using comorbidities to predict the occurrence of NI in elderly patients with hip fracture, more attention should be paid to the common diseases of the elderly than to the serious diseases, which are often thought to make stronger contributions to mortality.
This study has a few limitations. For example, it was retrospective with a long follow-up period (10 years). In addition, the records of inpatients came from different hospitals. Although stratification was used to randomly draw control cases, and the hospital from which the cases came was used as the basis for stratification, a heterogeneity of cases was also unavoidable. Therefore, future research designed to take into account the economic, medical conditions, and demographic characteristics of different regions is needed to provide stronger clinical evidence.
Conclusion
Considering their associated high economic costs and somber prognoses, nosocomial infections in elderly patients with hip fractures warrant serious consideration among clinical surgeons and further research. In this study, age, gender, and the number of comorbidities were found to be associated with the occurrence of NI. In particular, the number of comorbidities was the most predictive factor for NI occurrence and was more accurate than CCI when used to build a regression model with age and gender to predict the occurrence of NI in elderly patients with hip fractures. Additionally, it is imperative to consider the common diseases of the elderly when investigating the relationship between comorbidities and NI in elderly patients.
Anzeige
Declarations
Ethics approval
This study was approved by the ethics committee of the Chinese PLA General Hospital. A written informed consent was obtained from all the patients.
Conflict of interest
The authors declare no competing interests.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Mit e.Med Orthopädie & Unfallchirurgie erhalten Sie Zugang zu CME-Fortbildungen der Fachgebiete, den Premium-Inhalten der dazugehörigen Fachzeitschriften, inklusive einer gedruckten Zeitschrift Ihrer Wahl.
Ein Frauenanteil von mindestens einem Drittel im ärztlichen Op.-Team war in einer großen retrospektiven Studie aus Kanada mit einer signifikanten Reduktion der postoperativen Morbidität assoziiert.
Gesundheitsminister Lauterbach hat die vom Bundeskabinett beschlossene Klinikreform verteidigt. Kritik an den Plänen kommt vom Marburger Bund. Und in den Ländern wird über den Gang zum Vermittlungsausschuss spekuliert.
Der Einsatz von Operationsrobotern für den Einbau von Totalendoprothesen des Kniegelenks hat die Präzision der Eingriffe erhöht. Für die postoperative Zufriedenheit der Patienten scheint das aber unerheblich zu sein, wie eine Studie zeigt.
Mit dem Hebelzeichen-Test lässt sich offenbar recht zuverlässig feststellen, ob ein vorderes Kreuzband gerissen ist. In einer Metaanalyse war die Vorhersagekraft vor allem bei positivem Testergebnis hoch.
Update Orthopädie und Unfallchirurgie
Bestellen Sie unseren Fach-Newsletter und bleiben Sie gut informiert.
