Short-term effects of COVID-19 on risk of traumatic fractures in China: a multi-city study

Introduction: Traumatic injury is a leading cause of death and disability worldwide, and fth most common of in China. Along with the outbreak of COVID-19, strict control measures to restrict people’s movement have been conducted in China. Subsequently, the injury mechanisms and pattern of traumatic fractures changed signicantly. This study aimed to investigate the associations between COVID-19 and fracture risk, and provide a targeted reference for the world through China’s experience. Methods: This was a retrospective study of a nationally representative sample of COVID-19 prevalence areas using stratied random sampling. The data of traumatic fracture sustaining patients, including age and sex, fractured sites, mechanism of injury, and concurrent fractures in selected hospitals, were collected from 10 January and 10 July, 2020. The epidemiologic characteristics of traumatic fractures and the associations between COVID-19 and fracture risk were explored using the descriptive epidemiological methods and distribution lag nonlinear model. Results: A total of 67,249 (52.3% males) patients (average age 49.4±19.4 years) with 68,989 fractures were included. The highest proportion of fractures were sustained to the tibia and bula (14.9%), followed by the femur (13.6%), and ulna and radius (12.5%). Low-energy fractures accounted for 23.3%. With the increase of newly conrmed COVID-19 cases, fracture risk decreased for children, young and middle-aged adults, elderly men, high-energy fracture, and for residents in low and middle-prevalence areas. Conclusion: Fracture risk decreased sharply in all residents except elderly women, low-energy fractures, and in high-prevalence areas when newly conrmed COVID-19 cases increased in China. Primary (home) prevention measures are emphasized to prevent traumatic fractures during the COVID-19 pandemic. of different genders in accordance with the normal distribution. ANOVA was used to compare age in different epidemic areas. Statistical analysis was performed using R 2.15.0 (R Foundation for Statistical Computing, Vienna, Austria). We studied the impact of COVID-19 on the number of fracture cases in a distributed lag model. The specic method is to establish a lagging cross-basis matrix of newly diagnosed COVID-19, add the natural cubic spline function of time a, and the week variable and holidays as covariates to control the confounding of trend and week effect.

The inclusion criteria were as follows: (1) de nite diagnosis of new-onset fracture; (2) the fracture was sustained between 10 January and 10 July, 2020. The exclusion criteria were (1) pathologic (metastatic) fracture, (2) secondary fracture of various causes, including poor union, nonunion or readmission, and periprosthetic fracture.

Fracture data collection and groups
All fractures in the selected hospitals from 10 January to 10 July 2020 were collected through the Picture Archiving and Communication System (PACS) and case reports checking systems. The collected data of interest included demographics (age and sex), the fracture site, the mechanism of injury, and concurrent fractures. All medical charts and radiographs for fracture patients in each participating hospital were evaluated by two local orthopedic surgeons and were addressed by discussion if there was any inconsistency. Based on age, all patients were divided into three groups: children (≤14 years), young and middleaged adults (15-64 years), and older patients (≥65 years). Subjects were also strati ed into the following age groups: 0-4, 5-14, followed by an age group of 10 years, and ≧65 years.
The fracture sites were recorded as proximal, shaft, and distal fracture for each limb long-bone (humerus, ulna and radius, femur, and tibia and bula); pelvic and acetabular fractures; scapula; clavicle; patella; spine; hand and wrist; foot and ankle; and other, including sternum, rib, and head fracture. Patients who met the following three criteria were considered to have sustained osteoporotic fracture: (1) occurred in one of the four sites (hip, thoracic and lumbar vertebra, distal radius, and proximal humerus), (2) ≥65 years old, and (3) low-energy injury. [12][13][14][15][16] The injury mechanism included low-energy and high-energy fractures. A low-energy fracture was de ned as a fracture caused by a fall from standing and low height (<1m). A high-energy fracture was de ned as a fracture caused by a tra c accident, blunt injury, sharp instrument injury, fall from high height, explosive crush injury, and others.

COVID-19 cases
The daily average COVID-19 cases comes from the National Health Commission of the people's Republic of China platform. The system ful lled the quality assurance and quality control tasks of the Chinese government. The monitoring station built by the National Health Commission of the people's Republic of China provides COVID-19 data to the system every day to continuously monitor the COVID-19 prevalence throughout China. From January 10 to July 10, 2020, which was the most serious covid-19 epidemic period, there were 183 newly diagnosed cases.bvcxz Management principle for traumatic fracture in the epidemic of COVID-19 To ensure that COVID-19 con rmed or suspected cases, or common fractures, were diagnosed in every patient, the recent epidemiological history of the patient was con rmed. At the same time, necessary laboratory and imaging examinations were performed, such as routine blood tests, C-reactive protein, erythrocyte sedimentation rate, lung CT, and viral nucleic acid detection of throat swabs and respiratory tract secretions. All patients with common fractures were treated with rst-class protective measures, strict preoperative preparation, intraoperative treatment principles, postoperative nursing and rehabilitation, and follow-up after discharge, so as to formulate a meticulous and operable plan. If COVID-19 was diagnosed or suspected, the isolation measures were immediately implemented and upgraded to level three protection. At the same time, the orthopedics department; infection department; respiratory department; intensive care unit; anesthesiology department; digestive department; and other relevant departments were immediately organized for consultation, and the precise treatment concept of "one person, one policy" was implemented. In principle, patients treatable with a brace were not operated on. For patients requiring surgery, we chose minimally invasive treatment under the premise of ensuring the quality of fracture reduction, so as to decrease the secondary impact of surgical trauma on the immunity of patients. According to the changes in the disease, the treatment plan was adjustable at any time.

Statistical analyses
Statistical analyses were conducted using the standard statistical software SPSS 23.0 (IBM, Armonk, New York, USA). The Kolmogorov Smirnov test was used to test whether the data conformed to the normal distribution. The t-test of two independent samples was used to compare the age of different genders in accordance with the normal distribution. ANOVA was used to compare age in different epidemic areas.
Statistical analysis was performed using R 2.15.0 (R Foundation for Statistical Computing, Vienna, Austria). We studied the impact of COVID-19 on the number of fracture cases in a distributed lag model. The speci c method is to establish a lagging cross-basis matrix of newly diagnosed COVID-19, add the natural cubic spline function of time a, and the week variable and holidays as covariates to control the confounding of trend and week effect.

Role of the funding source
The funding source has no role in study design, conduction, data collection or statistical analysis.
The corresponding author had full access to all the data in the study and had nal responsibility for the decision to submit for publication.

Results
Descriptive statistics of data During the study timeframe, there were a total of 67,249 patients with 68,989 fractures, including 65,715 (95.3%) patients each having one fracture; 1,377 (20%) patients with two concurrent fractures; 125 (2%) patients with three concurrent fractures; 18 (0.2%) patients with four concurrent fractures; 11 (0.1%) patients with ve concurrent fractures, and three (0.4%) patients with six concurrent fractures. In the Wuhan Union Hospital, there were two fracture patients con rmed with COVID-19 on March 1, and one patient was con rmed with COVID-19 on February 1, including one with femoral neck fracture, one with calcaneal fracture, and one with lumbar fracture.
There were a total of 84,622 newly con rmed COVID-19 patients from January 10 to July 10, 2020. The number of newly con rmed COVID-19 cases was the highest in February (69400) (Fig. 2a), While that of fracture cases was the most in April (479) (Fig. 2b).
Age-and gender-speci c characteristics There were 35,176 (52.3%) male and 32,073 (47.7%) female patients, with an average age of 49.4±19.4 years (range, 1-108). The age of males was 43.9±20.6 years, signi cantly younger than that of females (55.4±21.0 years; t=-71.434, P<0.01). This study collected data on 6,196 (9.2%) children, 42,800 (63.6%) young and middle-aged adults, and 18,235 (27.1%) elderly patients. The ratio of males to females was 1.9, 1.4, and 0.5 in each age group, respectively. It can be seen from the age-sex pyramid of patient composition that the majority of patients are over 65 years old females (Fig 3).

Prevalence area characteristics
The age in the middle epidemic area was 52.1±18.9 years, signi cantly older than that in the low (47.0±22.1) and high epidemic areas (46.4±23.9, F=580.839, P<0.01). There are more males than females in the high (1025/760) and low (16528/13553) epidemic areas, while which was almost the same in the middle (17623/17760) epidemic areas. Adults accounted for the highest proportion in high, middle and low prevalence areas, but the proportion of the elderly in low (Fig 4a) prevalence areas was less than that in high (Fig 4b) and middle prevalence areas (Fig 4c).

The correlation between variables
All fracture cases Associations between COVID-19 and fracture risk are presented as three-dimensional graphs and two-dimensional contour plots in Fig 7. The varying range of newly con rmed cases is 0-15,152, and the lag interval is 0-10 days. Fig. 7A shows that with the increase in the number of newly con rmed COVID-19 cases, the fracture risk rst increased and then decreased. As shown in Table 1, on the 0-7 days of lag, with the increase in the number of newly con rmed COVID-19 cases, the fracture risk increased rst and then decreased (P<0.05). On the eighth to tenth day of lag, with the increase in the number of newly con rmed COVID-19 cases, the fracture risk did not change (P>0.05). The contour plot shows a pro le of RR of newly con rmed cases with different concentrations and lag days, which makes the relationship between the three more intuitive. Through the "slice" diagram of the RR, it shows that with the new con rmed COVID-19 cases the risk of fracture is gradually reduced (Fig. 7B). Fig 7C provides a two-dimensional exposure lag response correlation. On day zero of lag, with the increase in the number of newly con rmed COVID-19 cases, the fracture risk increased rst and then decreased. On the tenth day of lag, with the increase in the number of newly con rmed COVID-19 cases, the fracture risk had no change. When the number of newly con rmed cases was 2,000, the fracture risk was higher at 0-10 days, reaching the peak on the fth day (RR=1.40, 95% CI: 1.00-1.26). When the number of new con rmed cases was 12,000, the fracture risk was lower at 0-10 days (RR<1).

Gender and age
The fracture risk for males and females decreased with the increase of newly con rmed COVID-19 cases (supplemental gure 1A-B). Supplemental g 1C-G shows that with the increase of newly con rmed COVID-19 cases, the fracture risk decreased for children, young and middle-aged males and females, and elderly men, while it showed little change for elderly women.
Prevalence area Supplementary Fig. 1H-J shows the impact of COVID-19 on the number of fracture cases in different prevalence areas. With the increase in newly con rmed COVID-19 cases, the fracture risk decreased in the middle-prevalence area and low-prevalence area on the 0-7 days of lag. With the increase in newly con rmed COVID-19 cases, the fracture risk had no change in the high-prevalence area.
Fracture location and injury mechanism Supplementary g. 1K-L shows that with the increase of newly con rmed COVID-19 cases the limb and trunk fracture risk decreased. Supplementary g. 1m-n shows that with the increase of newly con rmed COVID-19 cases, the single fracture risk decreased, while multiple fractures had no change. Supplementary g 1O-P shows that with the increase of newly con rmed COVID-19 cases, the high-energy fracture risk decreased signi cantly, while the decrease in lowenergy fracture risk was not obvious.

Discussion
The present study found direct evidence of the effect of COVID-19 on fracture cases in China. This study compared the data on fractures of 18 hospitals in China within half a year of the COVID-19 outbreak, showing that the relationship between COVID-19 and fracture cases demonstrated a clear S-shaped curve; with the increase of newly con rmed COVID-19 cases, the fracture risk of male, female, children, young men, young women, and elderly men decreased fracture risk in middle and low-prevalence areas, and signi cantly reduced the risk of limb and trunk fracture. These in uences appeared in the 0-7 days of lag, and no longer existed after the eighth day.
The risk of fracture could be in uenced by the pandemic of COVID-19. During the epidemic period, the government took a series of measures to limit travel and reduce the population ow. In many Chinese cities, supermarkets and shopping malls carried less than 5% of patron volume during the non-epidemic period on January 23, 2020 (https://baijiahao.baidu.com/s?id=1659210676897303602&wfr=spider&f or = pc). Since February 19, almost all residents were isolated in their own homes except for the staff, doctors, and administrators engaged in prevention work, until the number of newly diagnosed patients was zero. Our ndings suggest that, with the increase of newly con rmed COVID-19 cases, the fracture risk of male, female, children, young men, young women, and elderly men was decreased. This is because, during the epidemic period, some industries were shut down and production was stopped, the risk of fracture in young and middle-aged people was reduced; the risk of fracture in children was also reduced after school suspension. However, due to the reduction of exercise and the change in sedentary lifestyle during the epidemic period, elderly people who usually stay at home are more likely to sustain a low-energy fracture, especially for elderly women with severe osteoporosis. The risk of fracture did not decrease with the aggravation of the epidemic situation. Therefore, in an epidemic period, we should consider taking targeted fracture prevention and control measures at home, especially for elderly women. These measures can include preventing insu cient lighting and uneven ground in living rooms and washrooms, wearing proper walking aids and antiskid shoes, and arranging furniture. In the same way, for patients with hypertension, diabetes, and other diseases, we should ensure the timely and rational use of medication, and reduce or eliminate the use of psychotropic drugs such as sleeping pills.
In this study, with the increase of newly con rmed COVID-19 cases, the fracture risk decreased in the middle and low-prevalence areas on 0-7 days of lag. However, with the increase in newly con rmed COVID-19 cases, the fracture risk had no change in the high-prevalence area. This indicates that, due to the control measures taken by the government such as tra c control and shutdown, the fracture risk of residents in middle and low-prevalence areas is signi cantly reduced. The number of patients in the high epidemic area is relatively large, and there is still more personnel ow due to the treatment, drug purchase, and care of family and friends. On the other hand, it also shows that control of personnel ow in high epidemic areas is not strict enough.
In this study, with the increase in the epidemic situation, the single, limb, and trunk fractures decreased obviously, while low-energy fractures did not decrease, accounting for 23.3% of all injury mechanisms. This is worthy of attention. Abundant low-energy fractures will not only increase the consumption of scarce medical resources, but also lead to disability, reduce the quality of life of patients, and even cause death. [17][18][19] Long-term medical care caused by fractures will increase the chance of COVID-19 infection in hospitals. 20-23 There is also increased evidence indicating that the incidence and mortality rates of elderly patients with COVID-19 are very high. [24][25][26] In addition, in the context of COVID-19, less physical labor, a sedentary lifestyle, panic, and depressed mental state increase the risk of low-energy fractures such as falls and slips. 27-28 Therefore, attention should be paid to the prevention of low-energy fracture in the epidemic period of COVID-19. To avoid the occurrence of low-energy fractures, anti-osteoporosis treatment should be actively undertaken, and calcium agents and active vitamin D3 should be timeously supplemented to promote the absorption of calcium. 29-31 We should simultaneously guide people on healthy home-based sports activities as well as the correct self-regulation of emotion and psychological state.
This study has several limitations. First, this study is a retrospective design, which inevitably leads to recall bias. However, the data were collected through the PACS and the case report examination systems. The main information is demographic (age and gender), fracture site, injury mechanism, and other basic information, so the recall bias of this study is very small. Second, the number of clinical diagnosis cases was included in the number of con rmed cases according to the novel coronavirus infection diagnosis and treatment plan (5th trial version) (http://www.gov.cn/zhengce/zhengceku/2020-02/05/content_5474791.htm) on February 12, 2020. This led to a sharp increase in the number of patients on that day, which affected the overall ow of the model. Third, hospitals rather than individuals were randomly selected using the sampling method, as directly selecting individuals randomly in each administrative village or neighborhood community using this method would not have been practical in China.

Declarations
Patient and public involvement The study was approved by the Institutional Review Board of the Third Hospital of Hebei Medical University. This study was a retrospective review of hospital records. Data entry and analysis do not display or use patients' personal identi cation (ID) information. Therefore, the requirement for informed consents from patients was waived.
Author Contributors YZ conceived the idea for the study and managed the project. YZ, HL, XZ, JW, ZH, WC designed the study. YZ, HL, XZ, JW, ZH, WC, HW, CL, WW conducted the survey across China and collected the data. YZ, HL, XZ, WC performed the statistical analyses and wrote the statistical analysis plan. HL and XZ wrote the manuscript. JW, ZH and WC prepared the gures and tables. YZ, HL, XZ, JW, ZH, WC revised the manuscript. All authors interpreted the data and contributed to preparation of the manuscript. All authors have approved the submitted version. HL and XZ contributed equally.

Declaration of interests
We declare no competing interests. All authors have read and contributed to the submitted manuscript and have no con ict of interest to declare.