Introduction

Since December 2019, the new coronavirus disease 2019 (COVID-19) has spread worldwide and led to successive confinements of populations. The COVID-19 disease stems from animal transmission but overall, several other emerging viral zoonoses already exist [1, 2], and that can possibly bring about similar global pandemics. A careful assessment of mortality is mandatory because lessons from the COVID-19 pandemic may be useful in case of a future new pandemic.

Regarding trauma, the epidemiology of trauma changed during confinements with the decrease in the incidence of road or sports accidents, but an increase in the number of household accidents and domestic violence-related injuries [3]. However, hip fractures are specific because they affect in-home fall [4] mechanisms in the elderly population with low energy. Hip fracture incidence was stable over time in level 1 trauma centers in China [4], Spain [5], and the United Kingdom [6, 7], and the specific pathway flow was established to continue the treatment of these fractures. The principle of these pathways is to separate patients with COVID-19 from those without to limit transmission and treat COVID-19 symptoms.

Elderly fragility and institutionalization increase the risk of COVID-19 in patients with hip fractures [8]. Having said that, it becomes interesting to assess whether the excess of mortality in institutions during the pandemic is due to COVID-19 infection or baseline frailty. In New York, Egol et al. found similar baseline comorbidities between patients with and without COVID-19 and the 30-day mortality was equivalent in patients without infection in 2020, in comparison with patients in the same centers in 2019 [9]. All in all, the increase in mortality among patients with hip fractures has seemed to be directly related to acute COVID-19 infection [4–7, 9–11] and not due to underlying comorbidities so far.

In France, the first patients with COVID-19 appeared at the end of January 2020, but few other cases were suspected without biological confirmation by the end of 2019. The first death due to COVID-19 in France occurred in mid-February, and the first clusters were identified in a military base in the north of Paris by the end of February. At this time, our hospital was the first reference center for COVID-19 in Paris with several department reorganizations. A first confinement started on March 17, 2020 and was extended until May 11, 2020. A total of 26,643 persons died of COVID19 during this first period.

In contrast to the data found in the literature, the incidence of hip fractures in our level 1 trauma center decreased during confinement. Furthermore, although our department is responsible for musculoskeletal care in an area of at least 410,000 inhabitants, the incidence of patients with COVID-19 infection and hip fracture was lower than of those from other countries. Our Department provides orthogeriatric care and this pathway was seriously disrupted during confinement because the two geriatricians who used to be fully committed in the orthopedic ward were transferred to another department. This is a relevant data because literature shows that orthogeriatric pathways are responsible for decreasing postoperative mortality [12–14]. Moreover, the benefits of orthogeriatric comprehensive care seem to be the most important for autonomous and healthy elderly patients [13]. In such a situation, regardless of their COVID-19 status, the disruption of our orthogeriatric chain likely affected the outcomes for all patients with hip fracture. Such an increase in mortality about disruption had already been described in other specialties.

For all these reasons, mortality outcomes should not be only evaluated in a confinement period cohort, comparing patients with or without COVID-19, because actual mortality should be minimized. A comparison with an anterior cohort from a similar period with exhaustive medical and social evaluation appears to be mandatory. There are still poor comparative data for mortality in literature with such a methodology and orthogeriatric center.

This study aimed to evaluate the 30-day mortality in 2019 and 2020, whereas the COVID-19 epidemic was fully active in France with confinement. The main hypothesis was that the recruitment periods contributed significantly to early mortality risk.

Methods

This work has been reported in line with the STROCSS criteria.

Results

Descriptive data

Table 1 details the demographic and medical characteristics of the participants. Two groups did not present any difference in age, sex, fracture type, Charlson Comorbidity Index, ADL, IADL, and dementia rate. Eight of 79 patients in 2020 and three of 102 patients in 2019 were nonoperatively treated (p = 0.09). The average Charlson Comorbidity Index and ADL score for nonoperatively treated patients were 7.7 and 4, respectively, versus 6 (p = 0.02) and 4.7 (p = 0.2), respectively, for others.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic and medical characteristic of the cohort.

https://doi.org/10.1371/journal.pone.0263680.t001

Comparative Kaplan Meier curves for the overall 30-day mortality, in 2019 and 2020.

In 2019, five patients came from other regions or countries. When considering the 93 patients in 2019 from our region, the average distance between hospital and place or habitation was 3.66 km, the average distance in 2020 was 3.5 km, and remoteness was not significantly different between the two periods.

The average delay of surgery was 3 days in 2019 and 1.8 days in 2020 (p = 0.0002).

Main results

A total of 42 patients died, the overall 30-day survival was 92% (95% CI, 88%–96%), and the overall 6-month survival was 78% (95% CI, 78%–89%). In 2019, three patients died before 30-day postoperatively, and survival was 97% (95% CI, 94%–100%). In 2020, 11 patients died before 30-day postoperatively, and the survival rate was 86% (95% CI, 79%–94%). The difference between the two periods was significant (HR = 5, 95%CI, 1.4–18.2, p = 0.013) (Fig 1). The 6-month survival was also lesser in 2020 compared to 2019 (HR = 2.3, 95%CI, 1.1–5, p = 0.023). The 1-year survival for patients of the 2019 group was 84% (95% CI, 78%–89%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Comparative Kaplan Meier curves for the overall 30-day mortality, in 2019 and 2020.

Significant difference in the 30-day mortality between 2019 and 2020.

https://doi.org/10.1371/journal.pone.0263680.g001

19 patients had COVID-19 in 2020, either in the same period as acquiring fracture or afterward. In 2020, COVID-19 was the cause of the deaths of four of 19 patients, and three patients within 30-day. All causes for the two periods are presented in Table 2. Number of deaths due to bed ridden tended to be higher in 2019 than 2020 (6.8% versus 1.3%, p = 0.14). Among the seven patients who were nonoperatively treated in 2020, five did not die before October 1, 2020, and one of three patients nonoperatively treated in 2019 was still alive in 2020.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Medical decompensations and cause of death.

https://doi.org/10.1371/journal.pone.0263680.t002

The Charlson comorbidity Index score, age, dementia, diagnosis, and delay of surgery did not significantly influence the mortality. The following factors were associated with the 30-day mortality in the unadjusted analyses: the sex (HR = 3.44, 95%CI, 1.2–10, p = 0.01) and the ADL score (HR = 0.69, 95%CI, 0.5–0.9, p = 0.006). The IADL variable was included in the multivariable model once imputed (Table 3). The COVID infection was also selected for the 30-day and 6-month multivariable models (HR = 2.5, 95%CI, 0.7–9, p = 0.15 at 30-day and HR = 2.2, 95%CI, 0.8–5.7, p = 0.1 at 6-month). In the multivariable Cox’s PH regression, the period was still significantly associated with the 30-day mortality (HR = 8.7, 95%CI, 2.15–35, p = 0.002) and COVID infection was not significant (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Unadjusted and multivariable-adjusted Cox’s proportional hazard regressions for the primary outcome 30-day mortality.

https://doi.org/10.1371/journal.pone.0263680.t003

The 6-month mortality was significantly different between 2019 and 2020 (HR = 3.4, 95%CI, 1.2–9.2, p = 0.01) and the COVID infection did not interact significantly (p = 0.5). The proportional hazard assumptions were all accepted (p>0.05).

Discussion

Fractures significantly decreased from 102 in 2019 to 79 in 2020 in our center. The present study mainly aimed to evaluate the extent to which the period modified mortality after upper femur fracture, in our orthopedic department. The main results highlight that the 2020 period was associated with an increase in the 30-day and 6-month mortality, although patients did not significantly have more comorbidities. Mortality over risk in 2020 remained true in the multivariable models, and COVID-19 infection did not significantly affect mortality. Also observed in 2020 was a non-significant increased trend for nonoperative treatment. The delay of surgery was significantly lesser in 2020.

To the best of our knowledge, independent of COVID-19 infection, this comparative series is the first to highlight an increase in mortality during the COVID-19 pandemic. The causes of death were mostly related to the systemic decompensation of underlying diseases. The total disruption of our comprehensive pathway with orthogeriatric care is our main hypothesis to explain this surge in mortality, independent of COVID-19. At the beginning of the pandemic, several publications proposed new organizations to manage COVID-19 patients and separate them from other emergencies [19–24]. After a few times, several surgical and medical specialties, such as bacteriologists, neurologists and oncologists, warned about the side effects of usual medical pathway disorganization [25–28].

Under normal circumstances, geriatricians are present in the care ward, and the very part of hip fractures are transferred here. Geriatrician assist surgeons with managing comorbidities and providing early treatment in case of perioperative complications. Geriatricians and surgeons also work together by examining all the patients who are hospitalized in the ward on a daily basis. Patients would be discharged from the department and placed in rehabilitation centers according to the complexity of the care they needed.

However, during the COVID peak period, geriatricians and surgeons were obliged to work in separate wards in order to participate to COVID infection care. The orthogeriatric ward closed and all trauma were treated at the hospital which normally receives polytraumatized patients, under normal conditions. As a result, daily visits with geriatricians were no longer possible and medical support was not offered automatically. The elective surgeries were canceled, every operating rooms were dedicated for emergencies so the delay between the admission and the operation decreased in 2020. In addition, the fact that many rehabilitation centers had been closed meant that patients were often discharged to non-specific geriatric reeducation centers, which did not necessarily have the resources needed to provide the level of care available previously. Nevertheless, the vast majority of the 30-day deaths in 2020 occurred in our orthopedic department, and the rehabilitation centers did not seem to be involved in the surge of mortality.

We identified four series on hip fractures comparing the 2020 period to previous periods, in which the crude 30-day mortality rates ranged from 2.2% to 8.6% but superior to 4% in most cases [9, 29–31]. In our series, the crude 30-day mortality rate in 2019 was 3 of 102 (2.9%), corresponding to a 3% mortality rate with Kaplan–Meier methods, which represents the lowest rate compared with that in other series. Furthermore, our crude 30-day rate is extremely close to that observed by Lau et al. with an orthogeriatric clinical pathway [12]. The 1 year crude mortality was also lower in 2019 than that in other orthogeriatric surveys [32]. For all these reasons, a significant early mortality observed in the 2020 period may correspond to cumulative effects of the loss of orthogeriatric care, which is responsible for underlying disease decompensations and COVID-19 infection, although this last reason did not show a significant level.

The main limitation of this series is the low number of patients with COVID-19 in 2020, leading to a lack of power to demonstrate the significant impact of COVID-19 on mortality. Based on Kayani et al.’s results [33], the 30-day mortality would increase from 10% to 30% in the case of COVID-19 infection in patients with hip fracture, indicating that our retrospective power would be approximately 50%. The Bonferroni method used to take into account the family-wise error exposed to an increase in the type two error risk but, in our case, the two p-value were less than 0.025. Thus, we maximized the chances to reduce the type one error without rejecting any hypothesis. Another limitation involves medical record filling that has poorer quality in 2020 than that in 2019, especially because of comprehensive care disruption. This loss in quality was responsible for more missing values regarding the Charlson Comorbidity Index, ADL, and IADL in 2020, compared with those in 2019. The Charlson Comorbidity Index can be easily reconstructed, whereas ADL and IADL cannot, and missing values were the main source of bias. Nevertheless, the results after the MICE approach were consistent with the primary ones, indicating that the findings were robust. Another limitation is the difference in the number of fractures between the two periods, which decreased from 102 to 79 without certified explanations. Several hypotheses stand out to explain such a difference: decrease in trauma incidence, migration of the population outside the urban area during confinement, unusual transfer of emergencies into private facilities, some patients with hip fracture could have die in their institution and have not been transferred to orthopedic departments.

Conclusion

This series supports all older cases highlighting the increase in mortality after hip fracture during the COVID-19 pandemic peak. However, new information is essential in the 30-day mortality due to underlying disease decompensation, whereas COVID-19 infection did not reach a significant and independent level. The main hypothesis is that orthogeriatric pathways were totally disrupted compared with those in 2019, and such comprehensive care was used to provide small 30-day and 1-year mortality baseline rates. We warn doctors and health authorities that they should do their best in order to maintain the usual clinical pathways in all circumstances in order to avoid excess of mortality among elderly trauma patients in case of a new pandemic breakthrough. A retrospective assessment of the human resources and organization should be launched to provide a road map of actions for the future.