Rates of adherence to cancer treatment guidelines in Australia and the factors associated with adherence: A systematic review

Adherence to cancer treatment clinical practice guidelines (CPGs) varies enormously across Australia, despite being associated with improved patient outcomes. This systematic review aims to characterize adherence rates to active‐cancer treatment CPGs in Australia and related factors to inform future implementation strategies. Five databases were systematically searched, abstracts were screened for eligibility, a full‐text review and critical appraisal of eligible studies performed, and data extracted. A narrative synthesis of factors associated with adherence was conducted, and the median adherence rates within cancer streams calculated. A total of 21,031 abstracts were identified. After duplicates were removed, abstracts screened, and full texts reviewed, 20 studies focused on adherence to active‐cancer treatment CPGs were included. Overall adherence rates ranged from 29% to 100%. Receipt of guideline recommended treatments was higher for patients who were younger (diffuse large B‐cell lymphoma [DLBCL], colorectal, lung, and breast cancer); female (breast and lung cancer), and male (DLBCL and colorectal cancer); never smokers (DLBCL and lung cancer); non‐Indigenous Australians (cervical and lung cancer); with less advanced stage disease (colorectal, lung, and cervical cancer), without comorbidities (DLBCL, colorectal, and lung cancer); with good‐excellent Eastern Cooperative Oncology Group performance status (lung cancer); living in moderately accessible places (colon cancer); and; treated in metropolitan facilities (DLBLC, breast and colon cancer). This review characterized active‐cancer treatment CPG adherence rates and associated factors in Australia. Future targeted CPG implementation strategies should account for these factors, to redress unwarranted variation particularly in vulnerable populations, and improve patient outcomes (Prospero number: CRD42020222962).


K E Y W O R D S
guideline adherence, medical oncology, practice guideline, radiation oncology, surgical oncology

INTRODUCTION
Adherence to clinical practice guideline (CPG) recommendations for cancer treatment has been associated with improved patient survival outcomes.This has previously been demonstrated across various cancers including sarcoma, 1 multiple myeloma 2 and cancers of the breast, [3][4][5][6] cervix, 7 lung, 8 head and neck, 9 and colon. 10Despite this, wide variation in practice patterns persist across cancer streams internationally, with adherence rates ranging from 54% to 77% in breast cancer, 11,12 35% to 56% in lung cancer, [13][14][15][16] 42% to 54% in cervical cancer, 7 22% to 85% in non-muscle invasive bladder cancer, 17,18 24% to 85% in ovarian cancer, 19,20 67% to 81% in prostate cancer, 21,22 and 36% to 96% in colon cancer. 23,24CPG nonadherence, specifically underutilization of guideline recommended treatment (GRT) has been identified as an issue in Australia, including radiotherapy (RT) treatment, [25][26][27][28] brachytherapy (BT), [29][30][31] chemotherapy (CTx), 27,[32][33][34][35] and endocrine therapy 27,36 for a broad range of cancers.Similarly, overutilization of treatments that are unnecessary or associated with harm leads to wasteful healthcare spending and increased burden on the healthcare system. 37plethora of variables influences adherence to cancer treatment CPGs, including factors related to CPG development.These include content and format, agreement with the underlying evidence, the applicability of GRT to individual patients, CPG currency, and prescriptiveness of the recommendations. 38,39Similarly, organizational and clinician factors, such as disciplinary preferences and biases, access to treatment options, clinical culture of peer review and multidisciplinary care coordination, and patient specific factors influence adherence. 38,39tient and health specific factors, such as older patient age and comorbidities, also influence adherence to CPGs across a variety of cancers, 40,41 including cancers of the breast, 3,12 lung, 42 colon, 10 and head and neck. 9Older and less healthy patients are underrepresented in clinical trials, which may reduce clinician confidence in the evidence underpinning some CPG recommendations and contribute to low rates of adherence. 3,38Eastern Cooperative Oncology Group (ECOG) Performance status, 12,14 cancer stage, 7,9,16 patient race, 13,43 and socioeconomic status (SES) 43 have also been associated with low rates of CPG adherence within specific cancer groups.
Barriers to effective CPG implementation are context specific, 44 and it is unknown whether the factors associated with cancer treatment CPG adherence are common across different cancers in the Australian setting.A better understanding of the impact of poor adherence across cancer streams in Australia, and the patterns of factors associated with CPG adherence (receipt of GRT), is needed to guide the implementation of tailored interventions.This knowledge will address the care gaps and distally contribute to reducing variation in treatment and patient outcomes across the cancer healthcare system.This review aims to (i) determine the rates of receipt of cancer GRT in Australia across various cancer streams, (ii) identify factors associated with cancer GRT in Australian studies, highlighting factors common across cancer streams, and (iii) examine whether receipt of cancer GRT impacts on patient outcomes.

METHODOLOGY
The review 45 was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (27 item checklist) 46 (see Additional file 1).

Search strategy, abstract, and full-text review
The search strategy was developed in consultation with a research librarian to search literature across five online databases (Medline, Embase, PsycInfo, Scopus, Web of Science core collection).The initial search was conducted by the lead author in August 2021 and updated in June 2022.Search terms are presented in Additional file 2. Abstracts identified by the search were collated in EndNote X9, and duplicates removed.Unique abstracts were then exported to Microsoft Excel V2208 and assessed for eligibility (see criteria below).
All titles and abstracts were reviewed in pairs, by the lead author (MB) and a second reviewer (FR, GA, YT, BNGE, KL, RCW, EA, BL, CYL, KC, DFP, LvB, KH, RC, SSO, RN, or PH).Titles and abstracts were assessed for eligibility against predefined criteria, and eligible publications were then selected for full review.The lead author reviewed all full texts, paired with the 17 other reviewers.The acceptability of the interrater reliability scores of reviewing pairs was calculated using Cohen's Kappa score. 47Disagreements on full texts were resolved through discussions with MB and GA.Reasons why publications were excluded at the full text stage are described in Figure 1.The reference lists of included studies were also searched for additional eligible studies.

Eligibility criteria for abstracts and full-text review
Studies were eligible for inclusion if they reported adherence rates to cancer treatment CPGs, using Australian data.For the purpose of this review, 'Guideline adherence' includes terms such as receipt of GRT or compliance, concordance, or adherence with a recommendation in a cancer CPG, protocol, or CPG-based quality indicator.All treatment types reported in studies were included.The review had no publication year limits.
Studies were excluded if they did not 1.include active-cancer treatment CPG adherence as a reported measure (e.g., studies focusing on adherence to CPGs for cancer If more than one publication described results from an eligible study, the publication which reported results most relevant to the review was included, or both were included if they reported separate treatments or different results.

Data extraction
Information on study characteristics (demographics including cancer type, number of participants, proportion of males/females, mean participant age, as well as study location, study length/year, study type/cohort design, eligibility criteria, data source/s, CPG/s cited, origin of CPG/s), and key findings such as adherence rates, factors associated with adherence and patient outcomes such as survival rates, were extracted by MB using a data collection template specifically developed for this study.

Study quality assessment
The Joanna Briggs Institute (JBI) checklist for prevalence studies 48 was used to appraise the quality of each study.The tool comprises nine items.Each study was independently assessed by two authors (MB and SS), with any discrepancies resolved through discussion with GA.

Data analysis
A narrative synthesis approach 49 was used to analyze CPG adherence rates and associated factors.The World Health Organization's (WHO) five dimensions of adherence framework (which includes patient factors, health condition factors, health care system and team factors, socioeconomic factors, as well as factors related to medical therapy) 50 guided the categorization of factors identified in this study.This framework has also been applied to CPG adherence in other health areas. 51,52Given the variety of cancers, subgroups, and treatment types and changes in CPGs over time, meta-analysis was not conducted. 53

Rates of receipt of cancer guideline recommended treatment (GRT)
All studies reported CPG adherence (receipt of cancer GRT) as a measure, with varying rates of adherence reported across the different cancer streams.The most notable feature is that the studies differed considerably in whether they examined adherence to a few indicators or many, and whether they stratified by subgroup or not.Adherence rates were also variable (Table 2).As a result of this heterogeneity, meta-analysis was not possible.This was the case even for the four of the lung cancer studies that looked at one CPG, because they looked at different patient subgroups.The CPGs referred to by each study are presented in Table 1, whereas the CPG recommendations are outlined in Table 2.
Factors associated with receipt of GRT have been categorized according to the WHO's five dimensions of adherence framework: patient factors, health condition factors, healthcare system and team factors, socioeconomic factors, and medical therapy factors 73 (Table 4 and Figure 3).

Patient factors
Patient factors (age, gender, smoking status, birthplace, and Indigenous status) were significantly associated with DLBCL, CRC, breast, lung, and cervical cancer GRT (Figure 3, Additional file 6).
Patient age was associated with GRT across 12 studies, with younger patients more likely to receive GRT in 1 breast cancer study, 66 1 DLBCL study, 69 4 lung cancer studies, 54,56,58 and 3 CRC studies. 62,63,65Patient gender was significantly associated with GRT in five studies.Female cases were significantly more likely to receive GRT in one breast cancer study, 68 and one lung cancer study, 57 whereas male DLBCL cases were more likely to receive GRT, 69 as were male stage C CRC cases, 63 and male high-risk rectal cancer cases GRT. 65-smokers, compared to people who never smoked, were less likely to receive GRT in one DLBCL study 69 and one lung cancer study. 57n-small cell lung cancer and cervical cancer cases who identified as Indigenous were significantly less likely to receive GRT in one study. 58

Health condition factors
Health condition factors (cancer stage, cancer site/histology, tumor size, node status, histologic grade, endocrine receptor [ER] status and Progesterone Receptor status, Chest Wall Separation Distance, Breslow thickness, distant tumors, comorbidities, ECOG status, weight loss, and prior cancer) were significantly associated with CRC, DLBCL, melanoma, breast, lung, and cervical cancer GRT (Figure 3, Additional file 6).More advanced stage cancer cases were significantly less likely to receive GRT in three lung cancer studies, 54,57,118 one cervical cancer study, 59 and one CRC study. 63Cases with comorbidities were significantly less likely to receive GRT in one DLBCL study, 69 two lung cancer studies, 57,58 and one CRC study. 63ECOG performance was significantly associated with GRT in three lung cancer studies.In two studies, cases with good-excellent ECOG performance status (0-1) were more likely to receive GRT than cases with borderline ECOG status (2), but less likely than cases with poor ECOG status (3-4). 54,118In another study, cases with excellent ECOG (0) status were more likely to receive GRT than cases with a poorer ECOG status of ≥1 (good, borderline, or poor). 57  Caseload was significantly associated with GRT in four studies, although there was no clear trend across cancers.In one melanoma study, cases treated by clinicians with lower caseload were significantly less likely to receive GRT, 71 and cervical cancer cases treated by gynecological BT departments with higher caseloads were more likely to receive GRT. 61 Lung cancer cases who were notified at higher volume hospitals were less likely to receive GRT, 57 and high-risk rectal cancer cases treated by surgeons with lower caseloads were more likely to receive GRT. 65 Rates of GRT increased over time in one breast cancer study, 66 one DLBCL, 69 and one lung cancer study. 57Rates of GRT (CTx) for stage C colon cancer cases decreased overtime in one CRC study.[Sx]) were found to be associated with GRT in one breast cancer study 67 and one CRC study. 65ng cancer cases from the fifth SES quintile group (the least disadvantaged) were more likely to receive GRT than cases from more disadvantaged areas. 57,118This trend was not seen in other cancer streams; DLBCL cases from SES quintile 4 were more likely to receive GRT than those from fifth quintile group, 69 whereas cases with stages B and C rectal cancers from lower SES quintile groups (first, second, and third quintile groups) were more likely to receive GRT than the least disadvantaged group (fifth quintile). 63age C colon cancer patients who lived in areas of moderate accessibility (where access to services is significantly limited) were more likely to receive GRT than those in highly accessible areas. 62east cancer cases who traveled greater distances to access treatment services (more than 50 km) were also more likely to receive GRT. 67 Conversely, cervical cancer cases who lived 5-10 km from their treatment facility were less likely to receive GRT than those living closer. 59e location of hospitals was significantly associated with GRT in three studies, with cases treated in rural facilities for breast cancer, 66 DLBCL, 69 or node-positive colon cancer, 65 being less likely to receive GRT (compared to those treated in metropolitan areas).

DISCUSSION
The implementation of evidence into practice is fundamental to enhance patient outcomes and efficiency of healthcare and research expenditure.This review demonstrates the large degree of adherence variability across active-cancer treatment CPG recommendations in Australia and characterizes the factors associated with GRT.GRT was associated with increased survival rates for patients with breast cancer, 66 DLBCL, 69 lung cancer, 54,118 cervical cancer, 59 and colon cancer. 62 the 20 studies included in this review, 15   39 This supports the need for further generation of real-world data 76 to build the evidence base to guide cancer treatments for older patients, those with later stage disease and comorbidities, in addition to the healthier and younger patients typically included in clinical trials. 38w SES was associated with poor GRT in lung cancer 57,118 ; however, the opposite was true for stages B and C rectal cancer patients, potentially as a result of patient preference, or clinical factors that influence treatment decisions. 63The association between low SES and poorer GRT for lung cancer is unsurprising given lower SES has been previously associated with limited access to curative Sx 77 and reduced cancer survival in Australia, despite universal healthcare coverage. 77,78tients with low SES also tend to have more comorbidities, 77 potentially limiting treatment options and CPG adherence.Poorer health literacy and education amongst lower SES groups also impact patient understanding of, and adherence to, cancer treatment. 79T was poorer for breast cancer cases, 66 DLBCL cases, 69 and CRC cases 65 treated in rural centers.Limited access to RT facilities as well as inadequate numbers of resident medical oncologists, radiation oncologists, and surgical oncologists may influence CPG adherence in non-metropolitan centers. 80Patients with low SES are disproportionately located in rural areas in Australia, 77 resulting in increased distance to travel (and associated time and costs) to access healthcare. 81Patient nonadherence as a result of travel burdens 39 may be addressed with further adoption of shared-care and telehealth technologies. 82,83GRT was higher for colon cancer patients who lived in moderately accessible compared to highly accessible areas, 62 and for breast cancer patients who traveled more than 50 km to access treatment, 67 possibly as a result of subsidized travel programs supporting such patients to travel to major referral hospitals for treatment. 62ven over a quarter of the Australian population (7 million people) live in rural and remote areas, 84 factors contributing to geographic variation in treatment and CPG adherence are important to consider.

Systematic considerations of equity during the development of
CPGs will contribute to ensuring treatments are equitably provided to disadvantaged groups, 85 reducing the variation in adherence rates identified across low SES, rural, and older populations.In addition, multifaceted implementation and dissemination strategies that target clinicians treating disadvantaged populations may reduce barriers to CPG adoption and adherence and enhance delivery of evidence-based practice across these groups.
Use of patient navigators 86 may encourage tailored treatment plans to support patients' individualized needs and have previously been identified as a facilitator of cancer CPG adherence in Australia. 39Other strategies that enhance CPG implementation by increasing clinician awareness and CPG uptake include education and opinion leaders, 87 CPG reminders, and audit and feedback of adherence rates, 88,89 particularly via Computerized Clinical Decision Support Systems (CDSSs) such as cancer therapy prescribing systems.In addition to enabling systematic audit and feedback processes, CDSSs are useful tools to support treatment decision-making and improve care and have been shown to improve adherence to cancer CPGs. 90Further integration of CDSSs into local Australian systems would better enable systematic audits to provide feedback to hospitals and clinicians.These tailored CPG implementation strategies need to be feasible, and acceptable for use in the target populations, to increase uptake of CPGs.
The findings from this review advocate for the collection and inclusion of real-world evidence that reflect the patient population 91 to support CPG recommendations that cater for a broader range of patient complexities.Implementation strategies should be tailored to specific populations that experience high rates of CPG nonadherence such as older, rural, and low SES populations and incorporate the wealth of implementation science knowledge as well as patient representation when developing, implementing, and disseminating cancer treatment CPGs.

Strengths and limitations
A strength of this review is the use of multiple reviewers to screen and assess studies, with generally strong interrater reliability scores.
The main challenge of such reviews is the inconsistent definition of CPG adherence across the Australian studies included, with studies reporting CPG adherence rates, nonadherence rates, compliance with quality indicators, and receipt of GRT.Multiple cancer streams were included in this review, with a small number of studies in each stream.This provides an overview of trends in adherence rates and factors associated with adherence, across streams; however, the heterogeneity of results across studies indicated that it was inappropriate to conduct a meta-analysis.It should be noted that the factors reported to be adherent with GRT are limited by the factors investigated in the included Australian studies and are not an exhaustive list of factors associated with GRT.While only studies published in English were included, this is to be expected when assessing Australian data.This study contributes to the literature by characterizing the rates of adherence to cancer treatment CPGs across Australia, and mapping the factors associated with adherence across a variety of cancer streams, enabling more tailored approaches to CPG implementation that will help to overcome barriers to uptake.

F I G U R E 1
Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) Study selection process.prevention, or the treatment of side effects such as pain management antiemetic prophylaxis, or palliation); 2. clearly report Australian patient care data (i.e., independently to data from other countries); 3. clearly define which CPG was being adhered to; 4. publish in English; 5. report quantitative and qualitative data separately; 6. report empirical research;7.include full texts (e.g., conference abstracts); and 8. publish in peer-reviewed journals (no gray literature was reviewed).

(
which were developed in partnership with the National Health and Medical Research Council (NHMRC)) (n = 7), National Comprehensive Cancer Network (NCCN) CPGs (n = 7), and NHMRC CPGs (n = 5) were the most commonly used CPGs across the 20 included studies (with 7 studies referring to more than 1 CPG) (

F I G U R E 2
Overall and subgroup cancer treatment clinical practice guideline (CPG) adherence rates by cancer stream in Australia.F I G U R E 3 Factors associated with guideline recommended treatment (GRT) across six cancer streams.

3. 5 . 3
Healthcare system and team factors Healthcare system and team factors (caseload, year of diagnosis, being discussed by a multidisciplinary team meeting [MDM], release of CPGs, treating clinician, surgeon specialty, referral physician conducting the excision, and clinician age) were associated with CRC, breast, melanoma, lung, and cervical cancer GRT (Figure 3, Additional file 6).

3. 5 . 4
Socioeconomic factors and Medical factors Socioeconomic factors (area-level socioeconomic disadvantage as measured by the Index of Relative Socioeconomic Disadvantage [IRSD], geographic remoteness of residence as measured by the Accessibility/Remoteness Index of Australia score [ARIA]/distance to treatment facility, treatment in a public or private hospital, hospital location [rural vs. metropolitan], area of residence and SES of the medical practice) were found to be significantly associated with CRC, melanoma, DLBCL, breast, lung, and cervical cancer GRT (Figure 3, Additional file 6).Medical Factors (treatment received, and curative intent of surgery

Table
).The data included in each study spanned the years from 1997 to 2018, with half of the studies reporting data from a time period that included 2006 (see Additional file 3).Three quarters of the studies were published after 2014 (see Additional file 4).Assessment using the JBI prevalence study critical appraisal tool demonstrated that the included studies were of high quality (see Additional file 5).

Study Cancer type and stage N of participants Study location Data year/study length Participant age and gender Study type Eligibility criteria Data source CPG CPG origin
Aust (Continues)TA B L E 1 (Continued)Study Cancer

stage N of participants Study location Data year/study length Participant age and gender Study type Eligibility criteria Data source CPG CPG origin
Aust Abbreviations: ACCPHSPC, American College of Chest Physicians, Health and Science Policy Committee; ACN, Australian Cancer Network; ACT, Australian Capital Territory; ANBCA, Australian National Breast Cancer Audit; ASTRO, American Society for Radiation Oncology; Aust, Australia; BCTG, Breast Cancer Treatment Group; BQA, BreastSurgANZ Quality Audit; BT, brachytherapy; CPG, clinical practice guideline; CRC, colorectal cancer; DLBCL, diffuse large B-cell lymphoma; EBC, early breast cancer; ESHL, early-stage Hodgkin lymphoma; ESMO, European Society of Medical Oncology; F, female; FIGO, International Federation of Obstetrics and Gynecology (Federation Internationale de Gynecologie et d'Obstetrique); GMCT, Greater Melbourne Clinical Taskforce; GOSG, NSW Gynecological Oncology Study Group; GRT, guideline recommended treatment; JSGO, Japan Society of Gynecologic Oncology; M, male; MDC, Multidisciplinary Committee; MDM, Multidisciplinary Meeting; NCCAC, National Collaborating Centre for Acute Care; NCCN, National Comprehensive Cancer Network; ND, no date; NHMRC, National Health and Medical Research Council; NSCLC, non-small cell lung cancer; NSW, New South Wales; QLD, Queensland; RCRCOIN, The Royal College of Radiologists Clinical Oncology Information Network; RT, radiotherapy; SA, South Australia; SACCR, South Australian Clinical Cancer registry; SACR, South Australian Cancer Registry; SCLC, non-small cell lung cancer; SE, South Eastern; SIGN, Scottish Intercollegiate Guidelines Network; SLHD, Sydney Local Health District; SWS, South Western Sydney; TNM, Tumor Node Metastasis staging system; VIC, Victoria.TA B L E 2 Overall reported adherence rates to active cancer clinical practice guideline (CPG) recommendations

Colorectal cancer Overall adherence rates Colon cancer, stages A, B, C, and D treatment CPG adherence rate
American College of Chest Physicians, Health and Science Policy Committee; ACN, Australian Cancer Network; Adj, adjuvant; ADT, androgen deprivation therapy; ASTRO, American Society for Radiation Oncology; BT, brachytherapy; Conv, conventional; CRC, colorectal cancer; CTx: chemotherapy; DLBCL, diffuse large B cell lymphoma; EBRT, external beam radiation therapy; ECOG, Eastern Cooperative Oncology Group; ER, endocrine receptor status; ESHL, early-stage Hodgkin lymphoma; ESMO, European Society of Medical Oncology; ET, endocrine therapy or hormonal therapy; FIGO, International Federation of Obstetrics and Gynecology (Federation Internationale de Gynecologie et d'Obstetrique; GMCT, Greater Melbourne Clinical Taskforce; GOSG, NSW Gynecological Oncology Study Group; Gy, gray unit; HDR, high dose-rate; HER2, human epidermal growth factor receptor 2; HF-WBRT, hypofractionated whole-breast radiation therapy; HR, hormone receptor; IMT, immunotherapy or immune targeted therapy; Interm, intermediate; JSGO, Japan Society of Gynecologic Oncology; LDR, low-dose-rate; MDM, multidisciplinary team meeting; MDT, multidisciplinary team; MRI, magnetic resonance imaging; NCCAC, National Collaborating Centre for Acute Care; NCCN, National Comprehensive Cancer Network; NHMRC, National Health and Medical Research Council; NSCLC, non-small cell lung cancer; NSW, New South Wales; OTT, overall treatment time; PL, pelvic lymphadenectomy; PR, progesterone receptor; RCRCOIN, The Royal College of Radiologists Clinical Oncology Information Network; RT, radiotherapy; SCLC, small cell lung cancer; SES, socioeconomic status; SIGN, Scottish Intercollegiate Guidelines Network; SLNB, sentinel lymph node biopsy; Sx, surgery; TME, total mesorectal excision; WLE, wide local excision; WPRT, whole pelvic RT. Median clinical practice guideline (CPG) adherence rates across cancer streams TA B L E 3Abbreviations: CRC, colorectal cancer; ESHL, early-stage Hodgkin lymphoma; HF-WBRT, hypofractionated whole-breast radiation therapy; NSCLC, nonsmall cell lung cancer.a One study reports adherence to lung and cervical cancer CPGs.
NS: not significantly associated with GRT; -: association between factor and GRT not reported.Abbreviations: ARIA, Accessibility/Remoteness Index of Australia; CPG, clinical practice guideline; CRC, colorectal cancer; DLBCL, diffuse large B-cell lymphoma; ECOG, Eastern Cooperative Oncology Group; ER, endocrine receptor status; IRSED, Index of Relative Socioeconomic Disadvantage; MDT, multidisciplinary team; SES, socioeconomic status; WHO, World Health Organization.