Introduction

The population health burden of osteoarthritis (OA) is significant and growing worldwide [1]. Elective hip and knee replacements are two of the most commonly performed operations in the UK to relieve the pain and functional limitations of OA (an indication for surgery in more than 90% cases) [2–4]. They are effective, with approximately 90% of patients achieving clinically meaningful improvements in pain and function [5], have very low rates of postoperative mortality [6, 7] and typical 10-year revision rates of less than 5% [2]. The mean age of patients at the time of their primary surgery is 68 years (hips) and 69 years (knees), and the proportion female is 60% for hip and 56% for knee replacements [2]. Primary hip and knee replacements use implants to replace diseased or worn out hip and knee joints. Revision operations are performed to add, remove or modify one or more components of a hip/knee replacement or to treat a periprosthetic infection. The lifetime risk of hip replacement is 11.6% for women and 7.1% for men, and for knee replacement 10.8% for women and 8.1% for men [8]. The mean length of hospital stay after primary hip and knee replacements is approximately three days [9], and is approximately 1–1.5 days longer for revision operations [10].

Seasonal spikes in hospital admissions, including for example a five percentage point increase in average bed occupancy rates in winter 2015/2016 compared with summer, have been a priority challenge for the National Health Service (NHS) for a number of years [11]. During peak demand in 2016/2017, one third of trust Accidents and Emergencies departments in England declared that hospital performance/patient safety may be compromised [12]. Between 2011/2012 and 2018, more than £3billion was provided at short notice to support NHS services over winter [13]. Management plans have been extended before the COVID-19 pandemic to include increasing bed occupancy above operating limits (85%) and cancelling elective operations [14].

In recent years, NHS England have advised wide-spread cancellations of elective operations on two occasions. The first was during the 2017/2018 winter flu season when NHS England advised hospitals to suspend elective operations throughout January 2018 [15]. This resulted in January-March 2018 recording the highest number of people waiting >18 weeks for their NHS treatment since quarterly records began (1994) [16]. The second was due to the Covid-19 pandemic. All elective operations in the NHS were suspended from 15^th April 2020 [17], initially for ‘at least three months’. In reality, many hospitals had begun cancelling these operations earlier in order to increase their capacity to treat adult COVID-19 patients [18].

The impact of the COVID-19 pandemic on the delivery of elective operations is still emerging as the pandemic continues to disrupt care. In this paper we describe the impact of the COVID-19 pandemic on elective hip and knee replacements at one teaching hospital in England.

Using data from one NHS Trust we will:

1. Compare the observed volume of elective primary and revision hip and knee replacements performed during the COVID-19 pandemic with those predicted from historic activity
2. Identify variation in the types of patients receiving elective primary and revision hip and knee replacements during the COVID-19 pandemic and any inequalities with respect to area-level deprivation

Methods

Results

Our study sample included 6,964 eligible operations performed between 1^st January 2016 and 30^th June 2021 at the Trust (S3 Table). Eight hundred and seventy-four (13%) were confirmed revision operations. In the ‘Pre COVID-19’ period 59% patients were female with a median age of 69. Fifty-five percent had no comorbidities and 13% had two or more. The most common comorbidities in this population were chronic pulmonary disease, diabetes without chronic complications, renal disease, and history of myocardial infarction (S4 Table). The proportion of patients increased as area-level deprivation decreased (most deprived 14%, least deprived 27%). The site of surgery was equally split between hips and knees.

A comparison of observed versus predicted hospital activity

Two hundred and thirty operations were performed during the ‘COVID-19 preparation’ phase, of which 90% were confirmed primary and 10% confirmed revision operations. Since the official suspension of elective operations (15^th April 2020) until the end of our study period, 690 operations had been performed (Fig 1), of which 85% were confirmed primary and 15% confirmed revision operations. Using time series modelling of surgical activity from January 2016 to December 2019, the predicted volume of elective primary and revision hip and knee replacements during COVID-19 preparation and restriction periods in the absence of restricted activity is 1,728 operations in total (Fig 2 and Table 1). The overall observed surgical volume was 53% of the predicted volume. The volume of knee replacement activity was more adversely affected by the COVID-19 restrictions than the volume of hip replacements, for both primary and revision operations. For primary joint replacements, the volume of knee replacements was 47.2% of predicted volume compared with 58.5% predicted volume for hip primaries. The volume of hip revision operations was 76.6% of predicted volume whereas knee revisions were only 41.7% of predicted volume. The predicted total bed occupancy during the COVID-19 preparation and restriction periods was 8,535 bed-days compared with observed occupancy of 4,393 bed-days (51.2%, Fig 3). For primary operations and hip revisions the proportions of observed to predicted bed occupancy were similar (primary hips: 56.9%, primary knees: 52.0%, revision hips: 57.9%), but were lower for revision knees (31.6%).

Download:

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

Fig 1. Annotated time-series of elective total hip replacements and knee replacements performed per month, 2016 to 2021.

Note–background colours signify phases of the COVID-19 pandemic: green = ‘Pre COVID-19’, orange = ‘COVID-19 preparation’, red = ‘COVID-19 restrictions’.

https://doi.org/10.1371/journal.pone.0294304.g001

Download:

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

Fig 2. A comparison of the observed volume of elective hip and knee replacements 2016–2019 (solid black) compared with the predicted volume (solid red) and observed volume (dashed black) between 1^st January 2020 and 30^th June 2021 per calendar month, by operation (primary or revision joint replacement operation) and location (hip or knee).

Note–background colours signify phases of the COVID-19 pandemic: green = ‘Pre COVID-19’, orange = ‘COVID-19 preparation’, red = ‘COVID-19 restrictions’.

https://doi.org/10.1371/journal.pone.0294304.g002

Download:

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

Fig 3. A comparison of the observed bed occupancy by patients receiving primary and revision elective hip and knee replacements 2016–2019 (solid black) compared with the predicted (solid red) and observed bed occupancy (dashed black) between 1^st January 2020 and 30^th June 2021 per calendar month, by operation (primary or revision joint replacement operation) and location (hip or knee).

Note–background colours signify phases of the COVID-19 pandemic: green = ‘Pre COVID-19’, orange = ‘COVID-19 preparation’, red = ‘COVID-19 restrictions’.

https://doi.org/10.1371/journal.pone.0294304.g003

Download:

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

Table 1. The observed volume of elective hip and knee replacements before (2016–2019 and 2018–2019) and during the COVID-19 pandemic (January 2020 to June 2021), and the predicted volume during the pandemic using 2 prediction models, by operation (primary or revision joint replacement operation) and location (hip or knee).

https://doi.org/10.1371/journal.pone.0294304.t001

Sensitivity analyses.

Predictions of surgical activity using a shorter pre-COVID time period (from January 2018 to December 2019) support our findings of substantially reduced observed surgical activity during than pandemic than predicted (Table 1). Predictions of primary joint replacement volume were higher than those from our main analyses, suggesting the impact of the pandemic restrictions could have been larger (lower proportion of observed to predicted activity). Predictions of revision joint replacement volume were more variable. The prediction of revision knee procedure volume using the shorter pre-COVID-19 period was similar to our main analyses. But the prediction of revision hip procedure volume was much lower using the shorter pre-COVID-19 period compared with main analyses.

A comparison of patients treated before and during the COVID-19 pandemic

In our comparison of patients treated between January 2016 and December 2019 (pre-COVID-19) with those treated during the COVID-19 preparation and restriction phases combined, we found a higher proportion of patients having a primary operation during the pandemic had 2 or more comorbidities than those treated before the pandemic (2 comorbidities: 13% versus 10%, 3 or more comorbidities: 4.4% versus 3.0%, P<0.001) (Table 2). The p-value was consistent with very strong evidence against the null hypothesis of no association. A higher proportion of patients having a revision operation during the pandemic had three or more comorbidities than those treated before the pandemic (3 or more comorbidities: 2.7% Pre COVID-19 versus 7.1% COVID-19), but the trend was not consistent for fewer comorbidities (No comorbidities: 53% Pre COVID-19 versus 51% COVID-19, P = 0.088). Revision operations were performed on younger patients during the pandemic than before (68 versus 72 years, P = 0.017), and the p-value was consistent with strong evidence against the null hypothesis. There were no differences in the sex distribution of patients treated during the pandemic compared with before the pandemic. A higher proportion of revisions were performed on people from the least deprived areas during the pandemic compared with before the pandemic (35% versus 27%) but otherwise differences with respect to IMD were inconsistent, with a large p-value for the comparison of primaries (0.720) and a p-value for the comparison of revisions (0.076) consistent with weak evidence against the null hypothesis.

Download:

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

Table 2. A comparison of patients treated by COVID-19 phase and operation type.

https://doi.org/10.1371/journal.pone.0294304.t002

Median length of hospital stay was shorter during the pandemic than before for both primary (three versus four days, P<0.001, Table 1 and Fig 4) and revision operations (five versus six days, P = 0.006). The adjusted Poisson regression models confirmed a reduced risk of longer hospital stays for patients treated during the pandemic compared with those before (incidence risk ratio [IRR] = 0.84; 95% CI: 0.81, 0.87; P<0.001 for primaries; IRR = 0.81;95% CI: 0.75, 0.86; P<0.001 for revisions). Comparisons using a shorter pre-COVID 19 time period (from 01/01/2018 onwards) showed similar results (S5 Table).

Download:

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

Fig 4. Median length of stay by operation type and site of surgery.

Note–background colours signify phases of the COVID-19 pandemic: green = ‘Pre COVID-19’, orange = ‘COVID-19 preparation’, red = ‘COVID-19 restrictions’; shading signifies the span from the lower (25%) to upper (75%) quartiles.

https://doi.org/10.1371/journal.pone.0294304.g004

A higher proportion of patients receiving primary or revision procedures during the pandemic had three or more comorbidities compared with before (primary: 3.7% versus 4.4%, revision: 3.3% versus 7.1% for Pre-COVID-19 and COID-19 respectively), but there was no clear pattern between the proportion of patient with up to two comorbidities and COVID-19 phase. The associated p-values were consequently large, providing no evidence against the null hypothesis. Post-hoc pairwise comparisons of the comorbidities of patients having primary joint replacements using chi-squared tests, corrected with the Benjamini and Hochberg method, indicated that for primaries no comorbidities versus any comorbidities were statistically significantly different (p-values<0.05, results not reported) whereas there was no difference between the groups with one, two or 3+ comorbidities. For revisions there was weak evidence of differences between three or more comorbidities compared with none, one or two comorbidities (p-values = 0.065, results not reported), with no evidence of differences between groups with two or fewer comorbidities.

Discussion

Patients who had an elective primary or revision hip or knee replacement between January 2020 and June 2021 at one large teaching hospital were younger than those treated before the COVID-19 pandemic, had more comorbidities, but also stayed in hospital for less time. We found no indication that the prioritisation of patients for surgery during the pandemic has introduced new inequalities in access to elective hip and knee replacements with respect to area-level deprivation. We found that the volume of elective hip and knee primary and revision replacements between January 2020 and June 2021 was 53% of the expected volume.

The reduction in volume of primary and revision hip and knee replacements at our study hospital is similar to the impact across England and Wales in 2020 compared with 2019 (56% of expected hips and 48% of expected knees) [29], and in other countries [30–32]. We have further highlighted a potentially concerning impact on revision operations as was observed in Poland [32], although we observed a greater impact on revisions of primary knee rather than hip replacements. Thaler and colleagues suggested that the reduced volume of revision procedures may be due to the lack of availability of intensive care units [33]. Although our predictions of expected revisions were more challenging than predictions of expected primaries, and findings may be less reliable. Primary joint replacements may need to be revised for different reasons, including for aseptic loosening, pain, dislocation, infection, and periprosthetic fracture [2]. Delays to revision operations longer than six months are likely to cause significant increases in pain and physical disability [34]. Revision operations are more complex than primary joint replacements and need a longer in-hospital recovery period [10]. Although these operations are much lower in volume, the reduced volume of revision operations may be of particular concern to Trusts if the trend we observed is replicated elsewhere.

We found no change in the area-level deprivation of patients treated during the pandemic compared with those treated before. This contrasts with a study at the same Trust of the impact of the suspension of elective operations during the 2017/2018 flu season, which found a reduction in the proportion of people from the most deprived areas having knee replacements [35]. To our knowledge there are no other studies of the impact of suspending hip and knee replacements on inequality in access to surgery. Patients treated during the pandemic were younger than those treated before the pandemic, in contrast to studies from other countries which found either a mixed effect of age depending on the joint being replaced [31] or no difference in age [36]. We also found that patients treated during the pandemic had more comorbidities. Younger patients in Spain who had their surgery re-scheduled in early 2020 were more willing to undergo surgery during the pandemic than older patients [37]. However, this did not extend to patients with comorbidities considered high risk with respect to possible complications from a COVID-19 infection. Further research regarding the epidemiology of hip and knee replacements during the pandemic may confirm whether our findings have been observed in other hospital Trusts.

We observed a reduction in the mean length of stay for primary and revision operations of one day during the pandemic compared with before. This is in agreement with findings from the USA [38, 39] and Poland [40] but contrasts with other findings from the UK [41]. This finding was consistent using either four years or two years of pre-pandemic data, suggesting this does not reflect a pre-pandemic trend. This reduction may be due to the selection of patients who were more likely to need a shorter length of stay, reflected in the younger patients treated during the pandemic. But the presence of comorbidities, which was higher for those treated during the pandemic, is associated with longer hospital stays [42]. We adjusted for patient factors, including number of comorbidities, which did not alter this finding. The shortened length of stay is unlikely to have been caused by pressure for beds from other hospital specialities but could be a consequence of infection control measures intended to minimise in-hospital transmission of COVID-19 infections. We do not know whether these patients had more post-operative complications or whether their recovery was adversely affected by their earlier discharge. If not, the safe introduction of shorter hospital stays could be a valuable finding to support the challenge of addressing the waiting list for elective hip and knee replacements.

Conclusions

Despite the significant impact of the pandemic on the provision of primary and elective hip and knee replacements, we found no evidence of new inequalities in access to these operations with respect to area-level deprivation. The shorter length of stay observed during the pandemic may inform changes to the postoperative management of patients and influence plans to address the waiting list for these elective operations. Further research looking at the impact of the pandemic on other elective operations and on the safety of shorter lengths of stay implemented during the pandemic may be valuable.

Supporting information

Acknowledgments

The authors are grateful to the members of the public who contributed through the patient and public involvement process and for Mike Bell and Carmel McGrath for organising the workshop.