Long-term follow-up of surgical complications and general health after preoperative optimisation for primary arthroplasty: a prospective quasi-experimental study

Abstract

Objectives

Preoperative optimisation programmes aim to reduce complications after total joint arthroplasty (TJA), yet long-term effects on superficial surgical site infections (SSI), periprosthetic joint infections (PJI) and overall health remain unclear. This study evaluated the long-term impact of preoperative optimisation in cooperation with hospital and primary healthcare on infection outcomes and general health following elective primary lower extremity TJA.

Design

In this prospective quasi-experimental study, 1484 patients undergoing TJA were followed postoperatively. The intervention group underwent a multifactorial optimisation process 6–12 months preoperatively, while the control group received conventional preparation.

Setting

Landspitali University Hospital and Primary Health Care in Iceland treating patients undergoing primary TJA between August 2018 and December 2022.

Participants

The intervention group comprised 746 patients and the control group 738. Of the intervention group, 77% received surgery during the COVID-19 pandemic versus 5% in the control group.

Outcome measures

Outcomes included incidence of superficial SSI at 1 year and PJI and postoperative health and mortality at 3 years.

Results

Superficial SSI occurred in 5.9% of the intervention group versus 7.9% in the control group (HR 0.67; 95% CI 0.45 to 1.01), with 95% versus 98% appearing within 6 weeks. PJI occurred in 2.0% versus 1.1%, respectively (HR 1.80; 95% CI 0.76 to 4.30). Most PJIs (80 vs 88%) followed superficial SSI, the majority within a week. Progression from superficial SSI to PJI was higher in the intervention group (27.3% vs 12.1%). No significant differences were found in postoperative health or mortality.

Conclusions

At 1 year postoperatively, no significant long-term reductions in superficial SSI were found, nor at 3 years in PJI or general health. Confounding factors, including COVID-19-related disruptions and organisational changes, may have influenced outcomes. The study emphasises the importance of prospective evaluation of preoperative optimisation strategies as well as organisational and practice changes on patient outcome.

Trial registration number

NCT05399186.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The study’s main strength is its prospective nature, detailed data collection, meticulous follow-up and minimal missing clinical data.

• The study groups were rather small but included exclusively patients undergoing primary total joint arthroplasty.

• Study progress, especially in the intervention arm, was affected by the COVID-19 pandemic and organisational changes during the study period.

Introduction

Lower extremity total joint arthroplasty (TJA) reduces pain and increases quality of life of individuals with end-stage osteoarthritis. The incidence of TJA operations continues to rise, with favourable outcomes in the majority of cases and prosthesis 10-year survival rate over 95%.¹ However, complications may follow TJA with superficial surgical site infection (SSI) estimated to occur in 1–10% of cases, with high risk of progression into periprosthetic joint infection (PJI).² This complication is devastating and costly for patients and society, requiring repeated surgeries and longstanding antibiotic treatment, and is associated with increased morbidity and mortality.¹ The incidence of PJI after TJA is between 0.5% and 2,8%, varying between countries.^{1 3}

Consequently, it has been pointed out that prevention of infections is more important than diagnoses and treatment⁴ leading to growing interest in preoperative optimisation programmes to improve modifiable risk factors before TJA to reduce complications and costs, and improve health outcomes after surgery.⁵

There is a wide variety in how preoperative optimisation has been approached. Organisation may be within hospital, primary care, special clinics or a combination.^6–8 The choice and number of health factors addressed differ and vary between medical, psychological and physical conditions of patients as well as socioeconomic status.^{5 9}

However, little has been published about the efficacy of such programmes^{10 11} and their possible long-term effect on general health.¹² We have previously reported an optimisation pathway, developed in collaboration of hospital and primary care, that increased patient engagement and improved modifiable risk factors preoperatively, leading to fewer surgical site complications and significantly reduced odds of developing superficial SSI within 6 weeks after surgery.¹³

The aim of this study was to prospectively estimate long-term infectious surgical outcomes and effect on general health for patients undergoing preoperative optimisation compared with conventional preparation.

Methods

This was a prospective quasi-experimental study conducted at a single institution. The control group¹⁴ and the effect of optimisation on the intervention group and their comparisons up to 6 weeks postoperatively have been described previously¹³ with methods, parameters and definitions used.

Patients in the control group were already registered on the TJA waiting list and enrolled between August 2018 and September 2020 during a preoperative clinic visit 1 week before surgery. They had received standard preoperative advice on weight, smoking and lifestyle from their orthopaedic surgeon at decision of operation without follow-up.

Patients in the intervention group were enrolled between January 2019 and January 2021 at the outpatient orthopaedic clinic after referral from primary care 6–12 months before surgery. Baseline blood tests were taken, and body mass index (BMI) recorded, with follow-up by their primary healthcare physician within 3 weeks, focusing on obesity, glycaemic control, anaemia, nutritional status, smoking cessation and exercise. Patients with BMI ≥40 were referred to a clinical nutritionist, and those with vitamin D <50 nmol/L were advised to take supplements. Malnutrition risk was assessed.

One week before surgery, both groups had blood tests and BMI recordings and completed a questionnaire regarding the preoperative preparation.

The control group had surgery from August 2018 to September 2020 and the intervention group from March 2019 to December 2022 and were then followed postoperatively.

Data registration included surgery details, demographics, anaesthesia type, American Society of Anesthesia (ASA) classification, laboratory variables, blood transfusions, length of stay, postoperative complications and mortality. Superficial SSI and PJI were diagnosed according to validated definitions.^{15 16} Details on joint-related complications postoperatively were recorded for those who required further surgery. A comparison of key variables between the two groups, which is provided in detail in a previous publication,¹³ is shown in online supplemental table S1.

Patient records from both primary care and hospital were manually reviewed to identify general health outcomes preoperatively and postoperatively to investigate the possible effects of contact with the healthcare system itself (including possible diagnosis of chronic diseases preoperatively and treating them unrelated to the optimisation programme) as well as the potential impact of the intervention on the occurrence of acute health-related events. General health outcomes were therefore divided into health-related diagnosis, reflecting appearance of chronic diseases, and health-related events, reflecting occurrences of acute conditions. It was considered more likely that preoperative optimisation would rather have an effect on diseases of an acute nature. Health-related diagnoses were defined as follows: cancer, hypertension, arrhythmia, ischaemic heart disease, heart failure, aortic stenosis, diabetes, obstructive lung disease, worsening of kidney function, increase in BMI and prescription of Glucagon-Like Peptide-1 (GLP-1) agonists. Health-related events were defined as follows: myocardial infarction, cardiac arrest, pneumonia, deep vein thrombosis, pulmonary embolism and transient ischaemic attack. Both health-related diagnoses and events were registered according to diagnosis codes in journals and records of prescription of GLP-1 agonists.

In the same purpose, information from hospital electronic health records was gathered including registrations of visits to emergency department, admission to hospital and admission to Intensive Care Unit (ICU) that occurred postoperatively as well as mortality.

The original study protocol is provided in the online supplemental materials, and a patient flowchart is shown in online supplemental figure S1.

Statistical analysis

The determination of group sizes has been described previously.¹³ Briefly, an initial analysis of the first 100 patients in the control group identified an 8% incidence of superficial SSI. Assuming that optimisation would reduce this incidence to 4%, a sample size of 550 patients per group was calculated as necessary to achieve adequate statistical power, although the original target enrolment was 1000 patients per group. Data are presented as frequencies and percentages for categorical variables and median (IQR) for numerical variables unless otherwise specified. Statistical significance was set at α=0.05.

Fine-Gray sub-distribution hazard models were used to analyse superficial SSI and PJI, treating death as a competing risk and adjusting for sex, age and ASA score. These factors were deemed most appropriate to adjust for, using ASA score as a composite factor reflecting the degree of underlying diseases. As a sensitivity analysis, we performed Cox proportional hazards regression model, which does not censor death, adjusting for the same clinical variables.

To study a possible influence of the COVID-19 pandemic on results, a sensitivity analysis using the Fine-Gray sub-distribution hazard model for superficial SSI was performed, adjusting the full cohort for COVID-19 as well as age, sex and ASA score. Also, subsets of the cohort before and after COVID-19 and during the same time period irrespective of COVID-19 were analysed adjusting for age, sex and ASA score. The two groups were also compared internally before and after COVID-19 without adjustments.

Cumulative incidence functions were estimated using Aalen-Johansen method, with Gray’s test comparing incidence curves across groups, accounting for death as a competing risk.

Univariate logistic regression was used to evaluate associations between modifiable risk factors (anaemia, HbA1c ≥42 mmol/mol, BMI ≥30 kg/m², smoking, vitamin D insufficiency, lack of exercise) and superficial SSI, stratified by treatment group. For count outcomes involving cumulative postoperative new illness diagnoses and health events, negative binomial regression was used to calculate incidence rate ratios with offset terms for variable follow-up duration. Both crude and adjusted models were fitted, controlling for age, sex and ASA score.

Kaplan-Meier survival analysis was performed for mortality outcomes, with log-rank tests for group comparisons.

Missing data were negligible for primary clinical variables.¹³ No imputation was performed. All analyses were performed using R V.4.5.0 (R Foundation for Statistical Computing, Vienna, Austria).

Patient and public involvement

Neither patients nor the public contributed to the design, analysis or reporting of this study. The results will be disseminated via peer-reviewed journals, conference presentations and as a part of a PhD thesis and dissertation.

Results

The control group consisted of 738 of 744 consenting patients included 1 week before operation between August 2018 and September 2020 (69% of eligible patients). The intervention arm consisted of 746 of 1010 consenting patients included at median 307 days (206–447), range 18–1322) before operation between January 2019–January 2021 (43% of eligible patients), with 259 exclusions due to declined surgery when offered. The control group had surgery between August 2018 and September 2020 and the intervention group between March 2019 and December 2022, or 5% versus 77% during the COVID-19 pandemic respectively. A comparison of key baseline variables between the two study groups is shown in online supplemental table S1.

Long-term follow-up after total joint arthroplasty

As the control group had their operations earlier in the observational period, there were a higher number of patients in the control group with follow-up over 3 years (online supplemental table S2). A cut was made at 1 year for analysis of superficial SSI and 3 years for other parameters.

Flow chart of long-term outcomes after operation for both groups regarding superficial SSI, Debridement, Antibiotics and Implant Retention operations, Two Stage operations and revision arthroplasties is shown in online supplemental figure S2. Revisions arthroplasties were more commonly performed in the intervention group, or 3.5% versus 1.9%.

Time factors

The median time from operation to diagnosis of the 44 superficial SSI in the intervention group was 21 days ((15–33), range 1–60), and of the 58 in the control group 18 days ((14–29), range 2–132).

The median time from operation to diagnosis of the 15 PJI in the intervention group was 31 days ((26–52), range 15–1056), and of the 8 in the control group 26 days ((22–36), range 17–512).

For those patients with superficial SSI that developed PJI the median time between the two diagnoses was for the 12 cases in the intervention group 6 days ((2–8), range 1–14), and for the 7 in the control group 4 days ((4–12), range 1–15) (figure 1).

Figure 1. (a) Time from surgery to diagnosis of superficial SSI after primary TJA (whole line and dot) and time to diagnosis of PJI after preceding superficial SSI (dotted line and cross). (b) Time from surgery to diagnosis of PJI after TJA without preceding superficial SSI (dotted line and cross). PJI, periprosthetic joint infection; SSI, surgical site infection; TJA, total joint arthroplasty.

HR for superficial surgical site infection

At 1 year, the occurrence of superficial SSI was less common in the intervention group or 5.9% veresus 7.9% in the control group. A Fine-Gray risk model adjusted for sex, age and ASA score showed a HR of 0.67 (95% CI 0.45 to 1.01) of developing superficial SSI in the intervention group, albeit the difference was not significant. A Cox risk model with adjustments for the same clinical variables showed a similar HR for occurrence of superficial SSI 0.67 (95% CI 0.45 to 1.00).

A crude estimate of the group effect on cumulative incidence of superficial SSI over 1 year postoperatively showed no significant difference between the intervention and control groups (Gray’s test p=0.132). When stratified by ASA score, significant differences were detected among the four curves (Gray’s test p=0.002). Similarly, when stratified by BMI categories, significant differences were seen among the four curves (Gray’s test p=0.041) (figure 2).

Figure 2. Cumulative incidence of superficial SSI (a) and PJI (b) by group, ASA and BMI categories at 1 versus 3 years’ follow-up time. ASA, American Society of Anesthesia classification; BMI, body mass index; PJI, periprosthetic joint infection; SSI, surgical site infection.

Sensitivity analyses on the possible effect of COVID-19 on the incidence of superficial SSI were all statistically non-significant (online supplemental table S3).

HR for periprosthetic joint infection

At 3 years, PJI was more common in the intervention group or 2.0% versus 1.1%. A Fine-Gray risk model adjusted for sex, age and ASA score showed a HR of 1.80 (95% CI 0.76 to 4.30) of developing PJI in the intervention group without significant difference between groups.

A crude estimate of the group effect on cumulative incidence of PJI over 3 years postoperatively showed no significant difference between the intervention and control groups (Gray’s test p=0.142). When stratified by ASA score, no significant differences were detected among the four curves (Gray’s test p=0.174). However, when stratified by BMI categories, significant differences were seen among the four curves (Gray’s test p=0.009) (figure 2).

Progression from superficial surgical site infection to periprosthetic joint infection

Superficial SSI preceded PJI in 80% (12/15) of cases in the intervention group and 88% (7/8) in the control group.

In the intervention group, 12 of 44 patients with superficial SSI progressed to PJI (27.3%) but in the control group 7 of 58 (12.1%).

The proportion of superficial SSI progressing to PJI was higher in both ASA 1–2 and ASA 3–4 categories in the intervention group compared with the control group. The highest proportion of progression was, however, in the intervention group among patients with BMI ≥30, or 37% (online supplemental table S4).

Association between preoperative risk factors and development of superficial surgical site infection

An univariable logistic regression analysis showed no association between superficial SSI at 1 year and the presence of known preoperative risk factors in the intervention group, but an association was found with elevated BMI ≥30 and HbA1c ≥42 mmol/mol in the control group (table 1).

Table 1. ORs for risk factors associated with superficial surgical site infection.

	Intervention group		Control group
	OR (95% CI)	Missing data	OR (95% CI)	Missing data
BMI at operation ≥30	1.31 (0.70 to 2.44)	0/746 (0.0%)	1.83 (1.04 to 3.21)	1/738 (0.1%)
HbA1c ≥42 mmol/mol	1.06 (0.48 to 2.35)	7/746 (0.9%)	2.09 (1.11 to 3.91)	17/738 (2.3%)
Preoperative anaemia	0.75 (0.18 to 3.20)	0/746 (0.0%)	0.89 (0.31 to 2.56)	2/738 (0.3%)
Vitamin D <50 nmol/L	1.14 (0.47 to 2.77)	5/746 (0.7%)	0.93 (0.44 to 1.94)	16/738 (2.2%)
Smoking	2.00 (0.58 to 6.92)	10/746 (1.3%)	1.19 (0.46 to 3.12)	11/738 (1.5%)
Weekly exercise ≤2	0.61 (0.29 to 1.26)	163/746 (21.8%)	0.92 (0.52 to 1.64)	71/738 (9.6%)

Missing data displayed as count/total (percentage).

BMI, body mass index.

Effect on postoperative general health

When looking at health-related diagnoses and events preoperatively, their total occurrence was higher in the intervention group than in the control group, reflected in higher ASA score in the intervention group.¹³ This pattern became the opposite postoperatively, both regarding new diagnosis and events (figure 3). A negative binomial regression analysis calculating incidence rate ratios (IRR), accounting for variable follow-up time and adjusting for age, sex and ASA score, showed no significant difference for IRR for in health-related diagnosis 0.79 (95% CI 0.61 to 1.02) nor in health-related events 0.94 (95% CI 0.59 to 1.49).

Figure 3. Number of all health-related diagnosis (reflecting chronic diseases) and health-related events (reflecting occurrences of acute conditions) in the control and intervention groups before and after total primary joint arthroplasty. TJA, total joint arthroplasty.

When analysing the cumulative incidence of first health-related diagnosis and event after operation there was no significant difference between the curves regarding first health-related diagnosis with lower incidence in the intervention group, or 16.6% versus 20.5%. Neither was there a significant difference regarding new health-related events, or 5.1% versus 5.3%.

There was a similar cumulative incidence between the intervention and control groups when comparing first visits after surgery to emergency department (46.0% vs 44.6%), admission to hospital (39.4% vs 41.6%) and intensive care (4.7% vs 4.5%) without significant difference between curves (figure 4).

Figure 4. Cumulative incidence of first health-related diagnosis and events, visits to emergency department and admission to hospital and ICU postoperatively. ER, Emergency Department; ICU, Intensive Care Unit.

Effect on mortality

There was no significant difference in survival curves between the treatment groups (log-rank test p=0.553), with 3-year survival at 98.1% in the intervention group versus 97.6% in the control group. When stratified by ASA score, there were significant differences among the four survival curves (log-rank test p<0.001), while no significant differences were detected among curves when stratified by BMI categories (log-rank test p=0.909) (figure 5).

Figure 5. Cumulative mortality according to groups, ASA and BMI score. ASA, American Society of Anesthesia classification; BMI, body mass index.

Types of bacteria

Positive cultures showed common bacteria known to cause PJI. Staphylococci were the cause of infection in the one case of PJI that occurred in the control group more than 6 weeks after operation but in 2 of the 4 that occurred in the intervention group while streptococci were the cause in the other 2 (online supplemental table S5).

Discussion

This study shows that even though preoperative optimisation improved modifiable risk factors and short-term incidence of superficial SSI and other outcomes, it did not decrease the incidence of superficial SSI significantly when registered 1 year after primary TJA. At 3 years, there was neither a difference in the occurrence of PJI nor a noticeable effect on general health of patients. However, differences between groups, the COVID-19 pandemic and substantial changes in organisation during the study may have affected results, emphasising the importance of following the outcome of patients on a continuous basis over time.

Short-term outcomes

The outcomes at 6 weeks have been previously reported showing that the optimisation effort increased patient engagement and improved modifiable risk factors before operation, leading to fewer surgical site complications and significantly reduced risk of superficial SSI but did not affect occurrence of PJI.¹³

Long-term outcomes

Extended follow-up of both superficial SSI and PJI was chosen to capture delayed presentation of superficial SSI and its possible progression to PJI. Superficial SSI typically presents early after TJA with a median time to diagnosis of 28 days.¹⁷ PJI is traditionally classified into early (≤3 months), delayed (3–24 months) and late (>24 months) infections.¹⁸ Early and delayed infections typically result from organisms inoculated during implantation that form biofilms on prosthetic surfaces, while late infections predominantly arise from haematogenous seeding or through compromised local tissues.¹⁹

This long-term follow-up also favoured evaluation of whether preoperative optimisation had any impact on general health after surgery.

Superficial surgical site infection

We chose superficial SSI as the primary outcome to increase the likelihood of detecting significant differences given the available sample size. This may be debated but is supported by evidence that superficial SSI often precedes PJI² as was the case in the vast majority in our study. The time frame for registering superficial SSI varies, from 30 days²⁰ to 6 months.²¹ We used a 1-year cut-off, with the last case diagnosed 132 days postoperatively. However, most cases (95% in the intervention group, 98% in the control group) were diagnosed within 6 weeks.

Reported superficial SSI rates after TJA range from 1 to 10%,² aligning with our findings: 5.9% in the intervention group and 7.9% in the control group, with a significant difference at 6 weeks,¹³ and overall similar rates at 1 year. Fine-Gray risk model adjusted for sex, age and ASA did not find a significant difference of developing superficial SSI between the groups at 1 year and a Cox risk model with the same adjustments confirmed the robustness of the findings.

When looking at cumulative incidence of superficial SSI, it appeared more associated with ASA score than BMI.

At 1 year, of the observed risk factors, elevated BMI and HbA1c were associated with superficial SSI only in the control group, the same pattern as at 6 weeks,¹³ possibly suggesting that optimisation may have mitigated risk in the intervention group despite higher comorbidities.

Sensitivity analyses did not reveal if COVID-19 pandemic had or had not effect on the incidence of superficial SSI between or within the groups.

Periprosthetic joint infection

At 3 years, PJI had occurred in 2.0% of cases in the intervention group and 1.1% in the control group, both within reported ranges (1–5%).² Most PJIs (80% in the intervention group, 88% in the control) developed from superficial SSI, typically within a week of the superficial SSI diagnoses, indicating infection burden was concentrated in the first 6 weeks postoperatively. A Fine-Gray risk model adjusted for sex, age and ASA score did not show significant differences between the groups. However, the study was underpowered to detect differences in this outcome.

Progression from superficial surgical site infection to periprosthetic joint infection

It was expected that superficial SSI was less common in the intervention group (significantly at 6 weeks but not in 1 year) but unexpected that PJI was more common (although not significantly). This appears to be explained by an also unexpected higher progression rate of superficial SSI to PJI in the intervention group (27.3%) than in the control group (12.1%), although the rate was similar in recent literature (29%).²

Factors influencing this progression remain unclear, but it likely depends on both patient characteristics and microorganism virulence.^{2 22}

The intervention group had a significantly higher ratio of patients with ASA score 3–4, or 32% versus 23%, but ASA score ≥3 has been significantly associated with such progression.²

The intervention group had a slightly higher ratio of patients with BMI ≥30, or 55% versus 52% but BMI ≥30 has been associated with increased risk of wound complications, superficial SSI and PJI following TJA.^{2 23} In our case, progression ratio was though highest among patients in the intervention group with BMI ≥30 (37%) and when looking at cumulative incidence BMI appeared more related to PJI risk than ASA score, though a recent study suggests ASA score may be a stronger predictor.²⁴

The ratio of diabetes was significantly higher in the intervention group or 12.5% versus 8.8%. However, diabetes has not been independently associated with progression from superficial SSI to PJI² but is identified as a risk factor for development of PJI.^{25 26}

The ratio of perioperative blood transfusion was significantly higher in the intervention group or 4.7% versus 2.6% but allogeneic blood transfusion has been associated with increased risk of both development of superficial SSI and PJI.²⁷

The above factors could all have contributed to an increased risk of progression from superficial SSI to PJI, or the development of PJI itself, possibly resulting in higher ratio of PJI in the intervention group compared with the control group. The higher comorbidity burden in the intervention group may have disproportionately impaired ability to contain superficial infections before extending to the prosthesis.

However, the significantly lower wound complication ratio in the intervention group or 11.3% versus 15.7% should have been protective both for development of superficial SSI and PJI.²

The intervention group was more affected by the COVID-19 pandemic with 77% of the group having surgery during it compared with 5% of the control group. The pandemic caused unprecedented disruption in regular healthcare services in Iceland as elsewhere, including delays of elective surgery, routine check-ups and effect on prioritisation of surgical procedures, staffing and preoperative and postoperative practices.²⁸ Social restrictions may have influenced the quality of follow-up as primary healthcare increased substantially telephone and web-based consultations during the pandemic.²⁹ The Healthcare in Iceland was though never overwhelmed by the situation³⁰ and the country experienced fewer deaths than expected during the COVID-19 pandemic resulting in the second-lowest excess mortality among OECD countries.³¹

However, reports on COVID-19’s effect on outcomes after TJA vary. Some report decline in SSI,³² others no change,³³ while some found increase in SSI,³⁴ especially in newly operated patients contracting COVID-19.³⁵

Increased susceptibility to infections in the initial post COVID-19 years have been reported.^{36 37} This may have affected the groups differently based on surgery timing, possibly influencing the progression from superficial SSI to PJI more within the intervention group. However, sensitivity analyses were inconclusive regarding the possible effect of COVID-19 on our results.

It is therefore difficult to account for the effect of COVID-19 pandemic possibly had on the study cohort.

Beyond this, the observed higher progression rate from superficial SSI to PJI in the intervention group should be interpreted with caution. The study was not powered to evaluate PJI nor progression from superficial SSI to PJI, and restricting the analysis to patients with superficial SSI would introduce post-treatment bias by conditioning on a mediator of the intervention effect. Therefore, the observed difference in progression is of uncertain significance and would require a larger, specifically designed study to determine whether it reflects a true effect.

Effect on postoperative general health

Preoperative optimisation has been shown to reduce infection rates,^{7 38} complications,³⁸ hospital stay,^{6 39} readmissions⁸ and costs⁴⁰ and improve patient engagement.⁴⁰

Our findings align somewhat, showing increased patient engagement and reduced short-term surgical complications and superficial SSI,¹³ though we observed no long-term effect on general health.

The intervention group initially had more patients with ASA scores of 3–4, reflected in a higher rate of preoperative health-related diagnoses and events.¹³ Postoperatively, this reversed, with more new diagnoses and events in the control group, possibly due to earlier detection in the intervention group from greater preoperative healthcare contact, though differences were not significant.

Postoperative cumulative rates of new diagnoses and events and first visits to emergency department, admission to hospital and ICU were similar between groups.

Mortality rates were also similar, but significantly higher in ASA 3–4 than ASA 1–2 groups but no such pattern was seen between BMI groups in line with a recent study.⁴¹

Types of bacteria

The most common causative organisms for PJI were gram-positive species, with staphylococci being the most common which is similar to other studies.²²

Limitations

The groups

The study groups were rather small and not fully comparable. The intervention group had more patients with comorbidities, reflected by a higher proportion with ASA scores of 3–4 (32% vs. 23%) and slightly more with BMI ≥30 (55% vs. 52%) despite health optimisation and weight loss.¹³ Heightened awareness and concern about superficial SSIs in the intervention group may also have led to more frequent diagnoses.

Patient selection

In 2021, due to long waiting times, Icelandic health officials arranged for TJA procedures at regional hospitals and a private clinic began offering TJA as well. Patients waiting over 3 months could also opt for surgery abroad under EEA/EU regulations. These factors likely contributed to a high number of patients declining surgery in the intervention group and a higher ASA score among those who participated.

Residual confounders

Given the differences between the groups and non-randomisation, we cannot exclude that the observed detected difference in outcome may be due to residual confounding. As noted above, several factors other than the intervention itself may have influenced the results

Strengths

The study’s main strength is its prospective nature, detailed data collection, thorough follow-up and minimal missing clinical data. We have also previously observed that the cohorts are comparable as described in similar studies^{13 14} supporting the generalisability of the results.

Conclusions

A preoperative optimisation effort of modifiable risk factors did not significantly reduce superficial SSI at 1 year or PJI at 3 years, nor did it have an effect on general health outcomes at 3 years. Changes during the study, particularly transfer of TJA procedures to other facilities and the COVID-19 pandemic may have influenced results by affecting patient selection and patient care practices.

This study reveals the importance of performing long-term prospective studies on preoperative optimisation as well as estimating the effect of organisational and practice changes on patient care.

Data availability statement

Data are available upon reasonable request.