Do Sustainable Palliative Single Fraction Radiotherapy Practices Proliferate or Perish 2 Years after a Knowledge Translation Campaign?

In early 2017, the Canadian Partnership Against Cancer and CancerCare Manitoba undertook a comprehensive knowledge translation (KT) campaign to improve the utilization of single fraction radiotherapy (SFRT) over multiple fraction radiotherapy (MFRT) for palliative management of bone metastases. The campaign significantly increased short-term SFRT utilization. We assess the time-dependent effects of KT-derived SFRT utilization 12–24 months removed from the KT campaign in a Provincial Cancer Program. This study identified patients receiving palliative radiotherapy for bone metastases in Manitoba in the 2018 calendar year using the provincial radiotherapy database. The proportion of patients treated with SFRT in 2018 was compared to 2017. Logistic regression analyses identified risk factors associated with MFRT receipt. In 2018, 1008 patients received palliative radiotherapy for bone metastasis, of which 63.3% received SFRT, a small overall increase in SFRT use over 2017 (59.1%). However, 41.1% of ROs demonstrated year-over-year decreases in SFRT utilization, indicative of a time-dependent loss of SFRT prescription habits derived from KT. Although SFRT use increased slightly overall in 2018, evidence of compliance fatigue was observed, suggestive of a time-perishing property of RO prescription behaviours derived from KT methodologies. Verification of the study’s findings in larger cohorts would be beneficial. These findings highlight the need for additional longitudinal KT reinforcement practices in the years following KT campaigns.


Background
There is significant evidence from high-quality published randomized clinical trials that single fraction radiotherapy (SFRT) is a more appropriate dose-fractionation choice when compared to multiple fraction radiotherapy (MFRT) for the palliative management of painful, uncomplicated bone metastases [1]. Several advantages of SFRT over MFRT we aimed to continue to quantify SFRT use in Manitoba in 2018 and assess the long-term sustainability of KT interventions in the radiation oncology milieu.
This study aimed to determine if KT-derived RO prescription behaviour for SFRT utilization declined in the long term (during the second year removed from the KT intervention), following the KT intervention. Specifically, we sought to determine if during 2018 there was any changes in the utilization of SFRT over the reported 2017 levels (59.1% for all bone metastases; 67.7% for uncomplicated bone metastases). We also sought to identify risk factors associated with receipt of MFRT during the same time period.

Methods
CCMB is the provincially mandated and publicly funded sole provider of RT services for the Canadian Province of Manitoba, with a catchment population of approximately 1.4 million persons.

Data Sources and Data Extraction
All courses of palliative RT for a bone metastasis in Manitoba during the study period (1 January 2018 to 31 December 2018) were identified using the CCMB radiotherapy database. This prospectively maintained electronic administrative database is populated with variables inputted into the RT treatment directive completed by a prescribing RO prior to the initiation of any RT-related treatment procedures. The following variables were extracted from this database for each course of RT: primary tumor type (ICD-10 diagnostic code), patient sex, patient age at time of RT, RT dose, RT fractionation, and prescribing RO. The remaining characteristics were extracted from the CCMB electronic medical records including anatomic treatment site, Eastern Cooperative Oncology Group (ECOG) performance status, radiotherapy (RT) treatment intent (post-operative vs. palliative; post-operative intent RT was defined as RT within 60 days after any orthopedic surgery intervention (e.g., open reduction, internal fixation surgery), and Charlson Comorbidity Index. Diagnostic imaging reports and the electronic medical records were used to classify each bone metastasis as complicated or uncomplicated by determining the presence or report of a fracture in the targeted bony structure, spinal cord compression, or cauda equina compression. Patients were excluded from analysis if the site receiving RT was predominantly a soft tissue metastasis where the bone metastasis was only a minor component of the target volume (defined as a bone metastasis that composes <20% of the target volume as determined using each patient's CT simulation scan). Patients treated with stereotactic body radiotherapy (SBRT) for bone metastases during 2018 were excluded from analysis since dose fractionation choices for SBRT patients are independent of the CWC guidelines.

Statistics
Baseline characteristics were tabulated for the entire cohort and by fractionation schedule (SFRT vs. MFRT). Differences in distribution of baseline characteristics by fractionation schedule were assessed using standard statistical tests (chi-squared, student t-test). The proportion of SFRT courses in 2018 was compared to the proportion of SFRT courses in 2017 using the one-sample z-test test for proportions. The proportion of SFRT courses prescribed by each individual RO was visualized with bar graphs for both uncomplicated and all bone metastases. The difference in proportion of bone metastases treated with SFRT were tabulated year-over-year for each individual oncologist, expressed as a percentage change. Baseline variables were assessed for potential associations for receipt of MFRT using univariable logistic regression analysis. A multivariable logistic regression model (Model 1) was built using the 2017 data employing a forward, stepwise approach. Variables with univariable associations of p ≤ 0.2 were considered for inclusion in the multivariable model and variables were assessed for collinearity in the model by assessing change in model variance during the forward stepwise selection process. A separate multivariable model (Model 2) was built including the data from the 2017 calendar year (previously reported) merged with the 2018 data with a variable added for year of treatment (2017 vs. 2018) in the model. The purpose of this separate model was to assess the odds ratio for receipt of MFRT by treatment year, adjusting for all the other potential confounding variables used in the 2017 logistic regression model. Multivariable associations with p ≤ 0.05 were considered statistically significant for this study. All analyses were conducted using STATA version 15 (Statacorp, College Station, TX, USA).
This study was conducted with the prior written approval of the University of Manitoba Health Research Ethics Board (Approval #: HS20808), and the CancerCare Manitoba Research Resource Impact Committee (Approval #: 2017-020).

Result
From 1 January 2018 to 31 December 2018, 1151 courses of palliative RT were administered to patients with a bone metastasis in Manitoba. Of these, 135 courses were excluded from the cohort because the metastasis was predominantly soft tissue metastasis with only a minor component of bony invasion. SBRT was utilized for 8 bone metastases, which were excluded from the analysis.
A total of 1008 courses of palliative RT were included in the analysis (Table 1) with a median age of 67 (range: 5-96), of whom 423 (42.0%) were women. The most common primary tumor types included: prostate (26.1%), lung (23.6%), and breast (17.3%). The most common anatomical sites of bone metastases included: skull/spine (44.6%), pelvis/proximal femur (32.3%), and upper extremity (9.2%). Retreatment to a previously irradiated site was done in 126 (12.5%) cases.   Table 2. For the group of nine ROs with increased SFRT utilization during 2018, who we will term "super adopters", the absolute percent change in SFRT utilization in 2018 over 2017 was +17.9%, with a range of absolute percentage increases of +2.2% to +47.1%. Conversely, for the group of eight with decreased utilization during 2018 (which we will term "lapsing adopters"), the absolute percent decline in SFRT utilization in 2018 over 2017 was −9.6%, with a range of absolute percentage change of −1.4% to −18.5%. Of the eight lapsing adopters, one RO decreased their SFRT utilization to below their baseline pre-KT campaign rate in 2016, while seven have maintained rates above their pre-campaign baseline but below levels seen in the year following the KT campaign (2017).
When restricting the analysis to those with uncomplicated bone metastases, the proportion of cases treated with SFRT in 2018 for uncomplicated bone metastases (72.0%) significantly increased over 2017 levels (67.7%; z-test p < 0.00001), representing an absolute year-over-year change of +4.3%. Among Manitoba's seventeen ROs present in the team during the 2017 calendar year, ten demonstrated increases year-over-year SFRT utilization for uncomplicated bone metastases, while seven demonstrated decreased year-over-year SFRT utilization ( Figure 2). Year-over-year change in SFRT utilization from 2016 to 2018 for uncomplicated bone metastases are described in Table 3. For the group of the ten super adopters, the mean absolute percentage increase in SFRT utilization in 2018 over 2017 was +19.5%, with a range of absolute percentage increases of +7.6% to +40.8%. Conversely, for the cohort of the seven lapsing adopters, the absolute percent decline in SFRT utilization in 2018 over 2017 was −10.5%, with a range of absolute percentage declines of −5.1% to −23.6%. Of the seven lapsing adopters, one RO had decreased SFRT utilization to below their baseline pre-KT campaign rate in 2016, while the other six maintained rates above their pre-campaign baseline but below levels seen in the year following the KT campaign (2017).
Curr. Oncol. 2022, 1, x 6 of 14 adopters, one RO decreased their SFRT utilization to below their baseline pre-KT campaign rate in 2016, while seven have maintained rates above their pre-campaign baseline but below levels seen in the year following the KT campaign (2017).     adopters, the mean absolute percentage increase in SFRT utilization in 2018 over 2017 was +19.5%, with a range of absolute percentage increases of +7.6% to +40.8%. Conversely, for the cohort of the seven lapsing adopters, the absolute percent decline in SFRT utilization in 2018 over 2017 was −10.5%, with a range of absolute percentage declines of −5.1% to −23.6%. Of the seven lapsing adopters, one RO had decreased SFRT utilization to below their baseline pre-KT campaign rate in 2016, while the other six maintained rates above their pre-campaign baseline but below levels seen in the year following the KT campaign (2017).    In 2018, MFRT was utilized for 370 (36.7%) of all bone metastases. For patients treated with MFRT, the most common fractionation schedule prescribed were: 20 Gy in 5 (77.6%) and 30 Gy in 10 (9.7%). The proportion of complicated bone metastases were similar between 2018 (31.7%) and 2017 (32.7%). Retreatment rates were also similar year-over-year (2018: 12.5%; 2017: 13.4%).

Discussion
This study found that radiation oncologist SFRT prescription behaviour imparted by KT campaigns demonstrated two important time-dependent characteristics in the second year removed from the KT campaign. Firstly, the rate of uptake in institution-wide SFRT utilization for all bone metastases has slowed from a 21.1% absolute increase in 2017 (0-12 months after the campaign) [19] to a 4.2% absolute increase in 2018 (12-24 months after the campaign; p = 0.0034), perhaps indicating that SFRT utilization is approaching its maximum asymptote or indicating the need for adapting our campaign message to continue reinforcing SFRT utilization. Secondly, a dichotomy has emerged whereby half of the ROs (9 of 17) who participated in the KT campaign continued to demonstrate year-over-year increases in their SFRT utilization for bone metastases in 2018, while the other half of the ROs (8 of 17) demonstrated a year-over-year decline in SFRT utilization in 2018 from their 2017 peak SFRT utilization rates. This is in contrast to the first 12 months after the campaign, when all 17 ROs who participated in the KT campaign demonstrated increased year-overyear SFRT utilization [19]. After controlling for potential confounding covariates included in Table S1, year of treatment (2018 versus 2017) remained as a statistically significant variable (p = 0.001). Thus, the positive impact of the KT campaign has not only decreased in momentum in the second year period post completion of the KT campaign but has also been carried by a smaller subgroup of ROs and some ROs have lapsed into old MFRT prescribing habits.
The decline in SFRT utilization observed in some of the ROs in our cohort two years post KT intervention may mirror the findings of a study conducted in British Columbia, Canada. In their jurisdiction. They examined SFRT utilization rate changes following a KT campaign in British Columbia, the authors also noted that SFRT use trended downwards after the initial uptick associated with their KT intervention [24]. The exact reason for these observed declines in SFRT utilization two years removed from the KT intervention both in the case of British Columbia and Manitoba are unknown but may be explained, in part, by several observations in the literature. Time-dependent loss of utilization of guideline compliant behaviour derived from KT campaigns has been observed in other KT milieus. In an observational audit study tracking hand hygiene compliance after a hand hygiene KT campaign in intensive care units, hand hygiene compliance increased immediately after the campaign, then subsequently declined during a two-year follow-up period as fewer intensive care units maintained strong compliance while other units returned to baseline lower compliance [25]. In another observational audit study tracking a hand hygiene KT campaign, initial hand hygiene compliance rose over the one-year period after starting the campaign, then decreased after the initial uptick [26]. These examples of time-dependent loss of KT derived behaviours suggest that the KT message compliance decays over time for healthcare practitioners and our findings suggest that radiation oncologists are not immune to forgetting lessons learned via KT campaigns. It is therefore a reasonable hypothesis that that re-exposing radiation oncologists to periodic KT refreshers may be helpful. It is also possible that lapsing radiation oncologists were overexposed to KT interventions leading to tuning out of the KT messaging, a phenomenon known as "messaging fatigue" or "campaign fatigue". Observations in the literature suggest that healthcare professionals who are regularly overexposed to KT campaigns are not immune to messaging fatigue [27]. Exposure of healthcare workers to excessive KT messaging has been associated with information overload and mental fatigue resulting in reduced ability to distinguish important messages from irrelevant ones [28,29] which in turn can lead to suboptimal care decisions and clinician behaviour [27]. In a randomized control study in Washington, the recall of a certain public health message sent to healthcare providers was inversely proportional to the mean number of messages received per week, and the odds of recall decreased with the increase of public health messages per week [30] suggesting that KT messaging faded into "background noise". Interestingly, that study was conducted during the H1N1 Influenza pandemic, which was associated with a dramatic increase in public health messages sent to healthcare providers, and recall rates improved as the overall message load on practitioners decreased to pre-pandemic levels [30]. More recently, evidence in the literature is emerging for the role of messaging fatigue in reducing participants' uptake of important healthcare information during the COVID-19 pandemic [31]. This corollary lesson from the H1N1 Influenza pandemic suggests that messaging fatigue is a potential outcome of overexposure to KT interventions, and one that administrators would need to keep in mind when determining how many interventions to expose radiation oncologists to with respect to SFRT use.
Although our study demonstrated the presence of a group of super adopters and another group of lapsing adopters of SFRT for bone metastases, this division into two subgroups of ROs did not exist in retrospect prior to the 2018 results. Neither subgroup, when analyzed retrospectively (data not shown), has consistently outperformed or underperformed the other subgroup in terms of SFRT utilization. Effective KT requires early addressal of barriers to knowledge adoption [16]. It is possible that the group of lapsing adopters may have encountered new and unique barriers to knowledge adoption pertaining to SFRT during year 2 of follow-up which may not have been addressed in the original intervention. Keeping these two subgroups in mind for future analyses of SFRT utilization in Manitoba may prove useful to identify and analyze barriers to resistance to the KT campaign messaging and improve upon the gains already achieved. Often, the drivers of maintenance behaviour in KT are different from the drivers of initiation [32], and successful campaigns must recognize this difference and adapt to a changing local context to remain relevant and sustainable over time [33].
Although the KT campaign's original goal was reaching an institution-level target of SFRT utilization for 60% of all bone metastases, as suggested by Tiwana et al. in a similar study in British Columbia [34], the results of this study suggest that this target may need to be revisited. Although the 2018 institution-level SFRT utilization for all bone metastases is 63.3%, we observed that only 72% of uncomplicated bone metastases received SFRT in 2018. Thus, there remains a considerable subgroup of uncomplicated bone metastases (approximately 20% of the whole cohort) for whom the SFRT guideline compliance can be improved. Moreover, there is growing evidence that SFRT is a clinically acceptable option to MFRT for subsets of patients with complicated bone metastases, specifically those with spinal cord compressions [35]. Thus, we judge that under ideal circumstances, the maximal proportion of all bone metastases which could and should be treated with SFRT in a guideline and evidence-complaint manner is approximately 80%. Our study identified several clinical subgroups of patients with lower proportions of SFRT compared to the rest of the population of patients with bone metastases, namely patients with RT prescribed post-operatively, hematological cancer primaries, non-prostate GU cancer primaries, and other cancer primaries (melanomas, head and neck primaries, gynecological primaries, sarcomas, primaries of the central nervous system, and unknown primaries). Future KT efforts will have to reaffirm the concept that SFRT for uncomplicated bone metastases is recommended for all clinical subgroups of patients including many patients from these aforementioned groups. To this end, we intend to continue to monitor compliance with SFRT utilization through audit and ad hoc feedback in an effort to keep the KT campaign message sustainable and relevant.
There are several limitations to this study. Firstly, the retrospective nature of our study cannot directly assess causality between the KT campaign and the observed effects on SFRT utilization. To mitigate this limitation, the combined 2017/2018 multivariable logistic regression analysis was built to assess the impact of treatment year on the outcome of receipt of SFRT while adjusting for many other potentially confounding variables in the model. Secondly, the decision regarding if a bone metastasis was complicated or uncomplicated prior to treatment was left to the discretion of individual ROs (i.e., it was not centrally reviewed or controlled), and thus may have had implications on the choice of SFRT vs. MFRT for individual ROs that was not captured when applying the definition of complicated employed in this study. Thirdly, since data regarding postoperative palliative RT was only collected in 2018, and not in 2017, we are unable to determine if any changes in proportion of patients treated postoperatively acted as a confounder in the year-over-year utilization of SFRT; however, this risk is expected to be minimal since postoperative RT consisted of only 5% of all bone metastases treated in 2018. Finally, although the scale of the original KT project was the full complement of ROs serving a catchment population of 1.4 million persons, this study was conducted on a total of 18 ROs. Thus, the specific proportion of super adopters and lapsing adopters seen in this study may differ if the study was repeated on a distinct population of ROs. For these reasons, further validation of our findings would be welcomed.

Conclusions
The rate of increase of SFRT utilization in Manitoba 2 years post KT intervention decreased compared to the immediate post KT-time period, and a significant proportion of ROs lapsed to lower SFRT utilization levels. Our findings suggest that KT-derived RO SFRT prescribing behaviour is time-perishing in nature. Further reinforcement of KT messaging and continued SFRT utilization audits are therefore warranted.