Prophylactic antibiotic regimens in tumour surgery (PARITY)

Study design and setting

We performed a pilot international multi-centre blinded parallel two-arm RCT. Each participating site obtained local institutional research ethics board approval and all patients provided informed consent. This trial was registered at ClinicalTrials.gov (NCT01479283) and is reported according to the Consolidated Standards of Reporting Trials (CONSORT) statement and recommendations for pilot studies.^13,14 We have previously reported our study protocol in further detail.¹²

Participants/study subjects

We included patients who were 12 years of age or older and had lower extremity primary bone malignancies, benign aggressive tumours, or soft-tissue sarcomas which had invaded bone and required surgical excision and endoprosthetic reconstruction.

We excluded patients if they had any of the following:

- Known methicillin-resistant Staphylococcus aureus or vancomycin-resistant enterococcus skin colonisation;

- Documented anaphylaxis or angioedema to cefazolin or penicillin;

- Prior surgery within the surgical field (excluding a biopsy);

- Prior local infection within the surgical field;

- Known immunologically-deficient conditions of disease such as HIV (not including recent chemotherapy);

- Known renal insufficiency with estimated creatinine clearance (eGRF) of < 54 mL/min;

- Reconstruction planned to include allograft (bone transplant);

- Likely problems, according to the judgment of the investigators, with maintaining follow-up;

- Unable to provide consent.

Randomisation

An unblinded pharmacist at each site randomised participants during surgery to either five days or 24 hours of post-operative prophylactic cefazolin. We used a central internet-based computer-generated randomisation system that concealed allocation. Our randomisation protocol included randomly permuted blocks of two or four and was stratified by location of tumour (femur vs tibia) and study centre.

The participants, surgeons, healthcare providers other than the pharmacists, data collectors, outcome assessors, data analysts, and those interpreting the results, were blinded to treatment allocation. Sites that used pre-mixed antibiotic bags kept them shrouded as there was a visible difference between pre-mixed bags and saline.

Interventions

Our protocol instructed that all participants would receive 2 g of IV cefazolin pre-operatively within 60 minutes of their procedures, and 2 g of IV cefazolin every three to four hours intra-operatively. Post-operatively, participants received either five days or 24 hours of cefazolin according to their randomised treatment assignments:

- Participants in the five-day arm received 2 g of IV cefazolin post-operatively every eight hours for five days.

- Participants in the 24-hour arm received 2 g of IV cefazolin post-operatively every eight hours for 24 hours, followed by sham doses of IV saline (placebo) for an additional four days.

- Paediatric participants received doses of cefazolin or sham that were based on 100 mg/kg/day doses, but were otherwise identical to the adult regimens.

Surgical excision and endoprosthetic reconstruction were performed according to the standard practice of the participating surgeons, which typically involved a wide extensile exposure, isolation and protection of major neurovascular structures, and resection of the segment of bone affected by tumour with a 2 cm to 3 cm bone margin. Decisions about implant selection, surgical techniques, soft-tissue reconstruction, or post-operative care other than antibiotic therapy, were left at the discretion of the treating surgeons, but were recorded in our Case Report Forms.

Outcome measures

In order to evaluate feasibility, we measured patient screening and enrolment, completeness of follow-up at each time point, and protocol deviations for the pre-, intra-, and post-operative antibiotic regimens.

The treating surgeons, study coordinators, and/or their delegates at each site identified potentially eligible patients at presentation and classified them as eligible and included, eligible but missed, or excluded. Completeness of follow-up was calculated relative to the number of patients eligible for follow-up at each time point.

We recorded protocol deviations whenever a pre-, intra-, or post-operative dose was missed, given incorrectly, or supplemented with additional non-study antibiotics. When patients were discharged before five days after surgery, we discontinued their study treatments early, recorded a protocol deviation for the missed doses, and no further antibiotics were given.

We pre-specified our criteria for success of the pilot as enrolment of our pilot sample within two years, 70% or greater protocol adherence, and 95% or greater completeness to follow-up.

To evaluate SSI rates, the participating surgeons diagnosed superficial, deep, or organ/space SSI according to the definition of the Centers for Disease Control and Prevention (CDC), which specifies that at least of one or the following criteria be met within one year after surgery:^12,15

- Purulent drainage from the incision;

- Organisms isolated from an aseptically-obtained culture of fluid or tissue from the incision;

- Incision deliberately opened by surgeon and culture positive;

- Incision deliberately opened by surgeon or designee and not cultured, butthe patient has at least one of: pain, tenderness, localised swelling, redness, or heat;

- Diagnosis of a superficial/deep/organ space incisional SSI by surgeon.

Monitoring

Before the start of patient screening and enrolment at each site, the Methods Center collected and reviewed the following documents:

- Site Principal Investigator’s current Curriculum Vitae;

- Physician’s Clinical Trial Application;

- Confirmation of ethics approval from the local institutional ethics committee;

- Approved informed consent form(s);

- Site delegation and signature log;

- Qualified Investigator Undertaking Form, if applicable;

- Research Ethics Board Attestation Form, if applicable;

Thereafter, the Methods Center ensured that the Site Principal Investigator, Research Coordinator, and Pharmacy Designee had received appropriate study-specific training, which may have included review of the training presentation during a teleconference call, independent review of the training presentation by the site personnel, or attendance at an Investigator/Coordinator meeting.

During the participant follow-up and clinical data collection phase of the study, the Methods Center remotely conducted the following ongoing monitoring activities:

- Review quality control reports from the iDataFax system to identify sites with unacceptable amounts of missing data or unresolved queries;

- Review enrolment reports to identify sites which have not been submitting screening data;

- Review the tracking database to identify any inconsistencies between the randomisation system and the submitted Case Report Forms with respect to treatment allocation;

- Review site-specific reports on study completion and loss to follow-up;

- Conduct periodic reviews of the site regulatory binders for missing and incomplete documentation;

- Review pharmacy logs.

After all subject follow-ups were completed at a clinical site, the Methods Center conducted the following remote closeout monitoring activities:

- Ensure that any missing data have been submitted (if available) and that all remaining data queries have been resolved;

- Ensure that all required adjudication materials have been submitted;

- Review the site regulatory file for completion and request any outstanding documentation;

- Review pharmacy logs;

- Request that the site close out the study with the local ethics board and submit a copy of the confirmation from the ethics committee to Chief Executive Officer for the site regulatory file

An independent Data Safety and Monitoring Board (DSMB) comprised of three orthopaedic surgeons and a statistician monitored participant safety. The DSMB reviewed adverse events and serious adverse events data on a quarterly basis, met via teleconference at least once a year, and was governed by a DSMB charter with terms of reference and functions.

Statistical analysis and study size

Baseline characteristics, feasibility data, and SSI rates are summarised as means with standard deviation (sd) or medians with interquartile range (IQR) for continuous variables and counts (%) for categorical variables. All analyses were conducted centrally at the Methods Center. Data analysis did not occur at any of the participating sites.

We enrolled a convenience sample of 60 participants to evaluate feasibility and we considered our pilot study completed when the last enrolled participant had achieved at least three months of follow-up. We estimated that 60 patients would represent approximately 5% to 10% of the definitive RCT sample size (600 to 1200 patients), based on a pairwise comparison with an alpha of 0.05, a beta of 0.20 (80% power), a relative risk reduction of 50%,⁸ and a range of plausible SSI rates.^12,16 We planned a priori to transition directly from our pilot RCT to a definitive RCT if feasibility was established, with the pilot event rate used to inform the definitive sample size estimation.

Recruitment

We screened 96 patients across 21 clinical sites in Canada, the United States, Australia, and Argentina between November 2012 and October 2014. Of these, 60 were eligible and included, none were eligible but missed, and 36 were excluded (Fig. 1). The number of participants enrolled at each site in this period ranged from zero (six sites) to 18 (one site), but the initiation of screening and enrolment was staggered across the participating sites because of variability in the time required to obtain ethics approval and negotiate institutional contracts (Table I). In total, 20 of the sites reported having a dedicated research nurse or research coordinator available to assist with the conduct of the trial, and 14 reported prior institutional research experience for clinical trials. Notwithstanding our inclusion and exclusion criteria, the mean estimated number of endoprosthestic reconstructions performed at each site per month was 1.8 (sd 1.7).

Fig. 1

Flow chart of participants in the trial.

By October 2014, 11 sites had been open for recruitment for 12 months or more, six sites had been open for six to 12 months, and four sites had been open for less than six months. The mean rate of enrolment in the trial after ‘enrolment-ready’ status was achieved was a mean of 2.13 (sd 2.4) participants per year across all sites.

Baseline characteristics

The baseline characteristics of the 60 participants are presented in Table II. There were 44 men and 16 women, and their mean age was 41.2 years (sd 23). The most common tumour location was femur (88%), and the most common tumour type was osteosarcoma (48%). In total, 47% underwent pre-operative chemotherapy, and 17% had metastatic disease at baseline.

Feasibility

As of 24 March 2015, 27 patients completed 12 months of follow-up, 23 patients had completed < 12 months but were still active in the trial, eight patients had died, one had withdrawn consent the day after surgery, and one had been lost to follow-up (Fig. 1). Of the patients that were eligible for each follow-up visit, complete data were collected for 98% at two weeks, 100% at six weeks, 86% at three months, 83% at six months, 90% at nine months, and 73% at one year (Table III). A further 17% of the patients eligible for follow-up at one year had partially complete data, and 10% had pending queries.

In total, 58 participants were randomised on the day of surgery, one was randomised on the first day after surgery, and one was randomised on the day before surgery. Protocol deviations occurred for the pre-, intra-, or post-operative antibiotic regimens of 37 participants (61%) (Table IV). Pre-operatively, three patients did not receive 2 g of IV cefazolin within 60 minutes of their procedures and three patients received other antibiotics in addition to cefazolin according to local institutional protocols, and in accordance with per-site stratification. Intra-operatively, seven patients received doses of IV cefazolin other than 2Â g every three to four hours owing to dosage adjustments for patient weight, and one patient received other antibiotics in addition to cefazolin.

Post-operatively, 18 participants missed at least one dose of cefazolin or placebo: 14 doses were missed among 11 participants because of pharmacy or nursing errors; 23 doses were missed among five participants because they were discharged before five days after surgery; 15 doses were missed in one participant who was started on alternative antibiotics after an intra-operative complication that led to a staged procedure; and 15 doses were missed in the participant who was randomised before surgery because they died before going to the operating room. The proportion of post-operative doses of cefazolin or placebo that were administered per protocol was 833 out of 900 (93%). Four patients received other antibiotics in addition to cefazolin post-operatively within the first five days after surgery.

Limitations

We required 24 months to enrol 60 patients across 21 sites from four countries. After accounting for variability in the timing of start-up at the clinical sites, our mean enrolment rate was 2.13 participants per site per year. Our screening data suggest that every eligible patient was enrolled (no patients were missed), but it is also possible that some eligible patients were not screened at all. For example, the most common type of tumour in our study was osteosarcoma, and the United States Cancer Statistics Working Group reported nearly equivalent incidence rates of osteosarcoma in men and women (5.0 per million vs 5.1 per million, respectively),¹⁰ however, we enrolled more men than women in our study, which raises the possibility of selection bias at the participating sites. Data from the Surveillance, Epidemiology, and End Results database, however, suggested higher incidence of osteosarcoma in men (5.4 per million vs 4.0 per million, respectively).¹⁷

Nevertheless, the most likely factor leading to the imbalance of male versus female participants is our small sample size and the subsequent likelihood of chance alone leading to the uneven gender distribution.¹⁸ This factor further highlights the critical importance of conducting large randomised trials as they ensure a balance of both known and unknown prognostic variables through the randomisation process.

Our recruitment data suggest that the most critical factor for the success of our definitive trial will be further expansion of our collaborative network. At the time of manuscript submission, the PARITY network consisted of 66 sites: 28 sites are open to enrolment, 30 sites are in various stages of ethics review or contracts negotiation, and eight further sites have expressed interest. With the addition of these sites to the collaborative network, we anticipate that the pace of enrolment will continue to accelerate.

The primary outcome for our definitive trial will be the rate of SSIs in each arm within one year after surgery, an endpoint that relies heavily on subjective clinical judgments.¹⁹ Although the surgeons in our pilot were blinded, it is possible our pilot event rate could be biased towards under reporting, given their involvement in the cases or that there could have been variability in the way that the CDC criteria were applied. In order to minimise bias and variability in the definitive PARITY trial, the definitive trial will implement a blinded Central Adjudication Committee in order to evaluate all potential SSIs according to pre-defined criteria.²⁰

Feasibility

Sarcoma patients require intense oncological follow-up to monitor for disease relapse, and we anticipated only minimal losses to follow-up in our pilot sample, but we also implemented several procedures in order to minimise losses.²¹ Although only 73% of the patients eligible for one-year follow-up had complete data available, we consider that our data support feasibility because there were only three patients with outstanding queries at 12 months, and a further 17% of eligible patients had at least partial data available. Given that some of the pilot patients are still within their one-year follow-up period, we expect to resolve most or all of these outstanding queries and missing data. Only one patient was lost to follow-up, although one additional patient withdrew consent on the first day after surgery. With respect to data quality in relation to participating site characteristics, we believe that in the pilot phase of this trial there was a learning curve for all sites, and reporting of data quality by site at this stage would be premature.

More than half of the participants experienced minor protocol deviations related to their peri-operative antibiotics, but the importance of these deviations to the validity and feasibility of a definitive trial is doubtful. Ten participants did not receive the correct regimens of pre- and intra-operative antibiotics, but it is implausible that these errors could introduce systematic bias, because allocation was concealed. Three participants received additional antibiotics pre-operatively according to local institutional protocols, but stratification should randomly distribute these deviations evenly between groups. Trials with protocol deviations can be considered within a ‘mechanistic-practical’ framework of design and interpretation, whereby mechanistictrials address the impact of interventions administered under ideal testing circumstances, and practicaltrials address the impact of interventions administered in ‘real world’ clinical practice.²² Given that most of the protocol deviations in our study reflected typical clinical practice, and that 93% of all post-operative doses were administered per protocol, these protocol deviations are unlikely to compromise the practical applicability of our results.

Furthermore, many previous trials in orthopaedic surgery have not adequately reported protocol deviations.²³ However, we reported according to the CONSORT statement and recommendations for pilot studies in order to improve the design and conduct of a subsequent large definitive trial.¹⁴ The doses missed because of pharmacy, nursing, or randomisation errors will be used as feedback to guide our future standard operating procedures.

We coordinated an expert panel of six orthopaedic oncology surgeons and three infectious disease specialists in preparation for this study.^8,12,24 Based on our survey data, expert opinion, and standard of care, we determined that the most appropriate antibiotic for this study was cefazolin. Our choice of five days as the long duration reflects consensus that an even longer duration would increase the risk for resistant organisms without providing further antimicrobial effectiveness.²⁵ We standardised the pre- and intra-operative antibiotic regimens in order to limit differential co-interventions.

Rates of SSI

In total, 15% of the participants in this study experienced a SSI. This rate exceeds the weighted mean of 9.5% identified in our systematic review (95% confidence interval (CI) 8.1 to 11.0), but this finding is not surprising, because most of the studies in our systematic review were retrospective case series at risk for selection bias or under reporting, owing to outcomes assessment bias. Prior studies have also used widely varying diagnostic criteria, and only two reported use of the CDC criteria.¹¹

The minimum follow-up in this pilot was just three months, but others havereported a median interval from implantation to infection of 8.5 months.²⁶ Therefore, it is possible that we could have detected an absolute risk of infection higher than 15% if all patients were followed up to one year. To the extent that this is true, a sample size calculation based on our pilot data might conservatively over power our definitive PARITY RCT, which would reduce the likelihood of reporting a spuriously negative result. Our pilot data support the idea that minimal or no adjustments to the definitive sample size calculation are required for losses to follow-up.

In conclusion, it is feasible to conduct a definitive multi-centre RCT of post-operative antibiotic regimens in patients with bone sarcomas, but further expansion of our collaborative network will be critical for study completion. We have demonstrated an ability to coordinate in multiple countries, enrol participants, maintain protocol adherence, and minimise losses to follow-up. The overall goal of PARITY is to provide high-quality evidence that can be used in the development of clinical guidelines. This pilot study has established a successful network and will support the rigorous design, organisation, and execution of a definitive RCT.