Measuring the quality of skin cancer management in primary care: A scoping review

Abstract Skin cancer is a growing global problem and a significant health and economic burden. Despite the practical necessity for skin cancer to be managed in primary care settings, little is known about how quality of care is or should be measured in this setting. This scoping review aimed to capture the breadth and range of contemporary evidence related to the measurement of quality in skin cancer management in primary care settings. Six databases were searched for relevant texts reporting on quality measurement in primary care skin cancer management. Data from 46 texts published since 2011 were extracted, and quality measures were catalogued according to the three domains of the Donabedian model of healthcare quality (structure, process and outcome). Quality measures within each domain were inductively analysed into 13 key emergent groups. These represented what were deemed to be the most relevant components of skin cancer management as related to structure, process or outcomes measurement. Four groups related to the structural elements of care provision (e.g. diagnostic tools and equipment), five related to the process of care delivery (e.g. diagnostic processes) and four related to the outcomes of care (e.g. poor treatment outcomes). A broad range of quality measures have been documented, based predominantly on articles using retrospective cohort designs; systematic reviews and randomised controlled trials were limited.


INTRODUCTION
Skin cancer is the most widespread form of cancer, with incidence rising worldwide. [1][2][3] The most frequently diagnosed skin cancers are non-melanoma skin cancers (NMSCs), mostly comprising the keratinocyte carcinomas (KCs), most of which are carcinomas of basal cells (BCCs) or squamous cells (SCCs). 4 Melanoma is a rarer form of skin cancer, affecting melanocytic cells, representing 1.7% of all cancers in 2020. 5 NMSC incidence is difficult to definitively determine because BCCs and SCCs are usually excluded from cancer registries. [6][7][8] The highest incidence of both melanoma and NMSC is observed in predominantly fair-skinned populations, such as those of Australia and New Zealand, 5,9 mostly due to high exposure to ultraviolet (UV) radiation from outdoor activities with insufficient sun protection. 10 In Australia, for example, melanoma is the third most common major malignancy after prostate and breast cancer. 11 NMSC is less likely to metastasise than melanoma, 12 but as it has 18-20 times the incidence, 10 NMSC and melanoma are both crucial parts of the skin cancer management challenge. 8,[13][14][15] For common cancers, primary care practitioners typically focus on prevention and diagnosis, and support patients while coordinating with specialists. 16 Many skin cancers, however, can potentially be managed entirely within the primary care setting [17][18][19] and, as incidence increases, demand for GP consultations and treatment for skin lesions has also risen. 20,21 There has been a lack of formal recognition and definition of the roles and responsibilities of general practitioners (GPs) in treating and managing skin cancer. 22,23 Research has drawn attention to GPs' capabilities in managing skin cancer but also to concerns around variation in the quality of care. [22][23][24][25] High levels of variability in diagnostic accuracy have been found between individual GPs, 26,27 and high variability in GPs familiarity with best practice guidance on high-risk excisions 28 and use of sentinel lymph node biopsy. 29 Skin cancer focused protocols and guidelines have been developed by dermatological and oncological societies (e.g. for surgical excision 30 ), but these have rarely detailed the role to be played by primary care. [31][32][33][34] GPs' approaches to skin cancer care have been found to be most influenced by their own training, interests, expertise and interactions with patients and colleagues. [35][36][37][38] Development of guidelines is insufficient to ensure high-quality care. Implementation of quality indicators, measurable elements of practice performance derived from guidelines, allow primary care practitioners to benchmark their performance against peers. [39][40][41][42][43] The Donabedian model of healthcare quality proposes that measures can relate to structure (i.e. attributes of settings), process (i.e. the giving and receiving of care) or outcome (i.e. effects of care on health status), with good structure and process contributing to better outcomes. 44,45 A set of quality indicators for the diagnosis and management of early stage cutaneous melanoma was recently developed, 46 targeting readily available measures of care processes such as pathology results. 46 It is also important to address the influence of setting (i.e. primary care) on the utility of quality indicators. 47 For example, is there a system in place to allow data to be understood and acted upon? Barriers to implementing quality measures differ across settings 42,48,49 and thus structural measures can affect clinicians' approaches to local quality improvement.
The aim of this scoping review was to better understand the literature on quality measurement of skin cancer management in primary care settings over the past decade. 44 Our approach was to keep the review broad, not limited to specific quality indicators that have been formally implemented or standardised, in order to understand the range and breadth of possible skin cancer care quality measures. Specific research questions relating to primary care skin cancer management were: 1. What types of evidence informs the measurement of quality? 2. What key groups of quality measurement have been explored or proposed?

MATERIALS AND METHODS
Relevant details relating to this study, and the project of which it is part, have been described elsewhere. 50 Selected details are described below.

Search strategy
A detailed search strategy was developed in association with an electronic information search expert (medical librarian) to optimise within each database the identification of relevant articles. 51,52 Six databases were searched on 1 December, 2021: Medline, PsycINFO, Embase, Scopus, CINAHL and Cochrane Library (see Appendix S1 for Medline search strategy). Searches were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) guidelines. 53 Where a selected article identified another article that contained relevant information, and the other article was also found within our initial six-database search but excluded during screening, that article was also included in the review. This restricted snowballing was used to protect against the inadvertent exclusion of relevant articles during screening.

Article selection
References were extracted into Endnote and duplicates identified and removed. References were uploaded into Covidence where titles and abstracts were screened by two team members (BIL screened all references, and the second reviewer was either LvB, DW, AEC, AS, CL, KH, MB, or FR) to assess compliance with inclusion and exclusion criteria (Table 1). If reviewers disagreed, a third reviewer (NS or LvB) facilitated consensus Full-text reviews were conducted by five team members (SS, NS, DW, BIL and LvB). Each article was independently read in full by two team members and assessed for eligibility. Disagreements were resolved through discussion; if needed, a third reviewer was consulted.

Data charting process
Data were extracted into a Microsoft Excel spreadsheet by two authors (SS and NS) and independently checked for accuracy (SS or GA). Extracted data were categorised as article details (authors, year, country, text type, objectives, conclusions, implications), study details (article type/study design, data source, setting, primary vs. secondary data, intervention type, control/comparison type), sample characteristics (type, size, attrition, gender, age, lesion type) and outcomes (type of quality measure, data source, indicator numerator and denominator). Where applicable, information about implementation was also extracted (acceptability, feasibility, reliability, validity).

Synthesis of results
Data from included articles were analysed by describing the breadth, range and type of included data and thematic analysis [54][55][56] to identify the underlying groups of quality proposed for measurement. Two team members (SS and NS) categorised measures according to the structure, process, outcome domains, 44 extracting data on a master sheet. SS and NS reviewed quality measures within each domain, discussed and generated a set of codes to represent the data, and summarised these codes into groups and subgroups of quality measurement. SS and NS met regularly to discuss discrepancies and reach consensus on categorisation and synthesis, consulting with GA regularly. Consensus-building teamwork during qualitative analysis helped confirm the trustworthiness of data and the veracity of resulting groups and subgroups. 57

Search results
As shown in Figure 1, 1315 references were identified, of which 353 were duplicates, leaving 962 articles for title and abstract screening. Of these, 740 did not meet eligibility criteria, leaving 222 articles for full-text review. After full-text review, 142 failed to meet the eligibility criteria (see reasons in Figure 1) leaving 80 articles. An additional seven articles were identified through snowballing. After removing 41 articles published before 2011, 46 articles were retained for review.

Characteristics of reviewed articles
The characteristics of included articles are displayed in Tables 2 and 3. Twenty articles (43%) were published from 2011 to 2016 and 26 (57%) were published from 2017 to 2022. Most articles were conducted and/or published in Europe (n = 29; 63%), particularly in the United Kingdom (n = 12; 41% of European articles). The rest came from North America (20%) and Australasia (17%). Six articles were practice guidelines or recommendations (13%), five were T A B L E 1 Inclusion and exclusion criteria for study selection in scoping review. systematic reviews (11%), one was a clinical literature review and one used a modified Delphi approach, with the remaining 33 having the following designs: retrospective cohort 21 ; cross-sectional, 7 two of which also had cohort elements; prospective cohort 4 ; and randomised controlled trial (RCT; 3).

Groups of quality measurement
Thirteen groups of quality measurement emerged through thematic analysis (see Table S1 for the authors that contributed to each group).

Structure measures of quality
Eighteen articles (39%) evaluated or proposed potential quality measures relating to structural elements of care provision; four groups of quality measures were derived (Table 4).

Diagnostic tools and equipment
Eight articles evaluated the effectiveness of diagnostic tools and equipment, falling within a single subgroup F I G U R E 1 PRISMA flowchart displaying the process of identification and selection of included articles. of inspection aids and imaging systems. These articles focused primarily on the use of dermoscopy and other diagnostic aids (e.g. MoleMate system), but also addressed image storage and retrieval platforms. [58][59][60][61][62][63] Practitioner education and training Six articles evaluated the impact of education and training programs on clinical practice. Most of these articles examined the effect of education and training for diagnostic tool-assisted skin inspections on detection accuracy, [59][60][61][62]64 while one sought to improve visual skin inspection. 65 Diagnostic protocols and documentation Thirteen articles assessed protocols and procedures to facilitate community or routine screening [65][66][67][68] or for the purpose of diagnosing suspicious lesions. [58][59][60][61][62][63][64]69,70 These articles recommended dermoscopy checklists and algorithms, [58][59][60][61] standardised recording forms [65][66][67] and visual skin examination checklists. 66 Treatment protocols and documentation Six articles 58,69-73 presented protocols and procedures for treatment, within a single subgroup of surgical and procedural safety. Recommendations included the use of guidelines for surgical safety, 58,69,70,73 surgical audit forms 71 and antibiotics use to prevent infection. 72

Process measures of quality
Forty-four articles (96%) evaluated or proposed potential quality measures relating to care provision, across five groups (Table 5).

Prevention
Three articles identified measures related to prevention. Behavioural counselling for younger patients was recommended as early prevention by US Preventative Services Task Force 74 and re-iterated. 66 Two guideline articles recommended high-risk surveillance practices including monitoring skin damage, UV light exposure and occupational risk factors. 69,73 Diagnostic processes Twenty-nine articles identified measures relevant to diagnosis-related processes of care, in four subgroups. These articles evaluated diagnostic accuracy relative to a T A B L E 3 Frequency of study characteristics included in scoping review.
T A B L E 5 Quality measures relating to processes of skin cancer care.

Group Subgroup Examples n (%)
Prevention Early prevention 74 Primary care-based counselling on ultraviolet exposure reduction for people aged 10-24 years with fair skin 74

Proportion of melanomas treated with imiquimod 79
Post-treatment follow-up 34,67,70,72,73,78,81,88,95 Proportion of melanoma patients requiring follow up after initial excision in primary care 81 Use of patient recall systems for each skin cancer type 95 Interpersonal process Communication with patient 58,88 Proportion of melanoma diagnoses communicated in-person, via phone and via post 88

(9)
Assessing patient care experience 64,88,96 Proportion of patients reporting satisfaction with melanoma care at post-surgery follow-up 88 Proportion of patient satisfaction surveys completed after lesion assessment within 1 week of consultation 64 Abbreviations: n, number of articles included in the thematic analysis; GP, general practitioner. (Continues)

Outcome measures of quality
Seventeen articles (37%) evaluated or proposed quality measures relating to outcomes of care, in four groups ( Table 6).

Treatment complications and adverse events
Six articles assessed treatment complications and adverse events such as post-operative infections, 35,71,91 as well as short-term morbidity indicated by post-treatment hospital admissions 85,86 and subsequent treatments. 67 Patient-reported measures Three articles evaluated PRMs, focused on patient satisfaction with care provided as cancer treatment 64,88 or patient-reported health outcomes such as anxiety or condition improvement. 64 One article reviewed implementation of patient-reported experience measures in practice. 96 T A B L E 6 Quality measures relating to outcomes of skin cancer care.

Treatment complications and adverse events
Post-operative infections 71,72,91 Proportion of surgeries for which infection occurred within 2 months 71 6 (13)

Rate of wound infections in patients with lower limb excisions 72
Short-term morbidity 67 Abbreviations: n, number of articles included in the thematic analysis; GP, general practitioner; PROMs, patient-reported outcome measures.
Post-treatment recurrence of skin cancer Six articles examined skin cancer recurrence rates, including NMSC recurrence after lesion excision 92 or suspected AK, 67 and melanoma recurrence post-melanoma surgery 78,79 or post-AK diagnosis. 67 Long-term morbidity and mortality Seven articles assessed long-term morbidity and mortality. Morbidity was measured as the proportion of cases that progressed to metastasis, 79,83,97 including from time of detection. 68 Mortality was measured as the proportion of cases that resulted in skin cancer death 80,83,85,97 or as a function of tumour thickness. 68,97

Types of articles
This scoping review identified 46 articles that suggest possible quality measures relevant to primary care skin cancer management, over the last decade. Most assessed skin cancer care quality through retrospective cohort articles, a design that provides valuable insights when RCTs are not feasible, 98 and a commonly employed to assess care quality. 99,100 Three RCTs assessed elements of care quality. 62,64,72 Five systematic reviews were identified, three with a meta-analytic component. 59,60,94 Quality measurement Thirteen groups of activities that may be suitable for quality measurement were derived. Most widely considered over the last decade are process measures, often referred to as 'intermediate outcomes' that provide actionable data on clinical and management processes in a timely manner, and thus are the most frequently utilised quality measures. 41,101,102 Five groups of process measures were identified: prevention, diagnostic process, delays in care, treatment process and interpersonal process. Diagnostic accuracy, a common focus, was assessed predominantly by comparing GPs diagnosis (either visually or tool-assisted) with histopathological 71 or dermatologist diagnosis. 75 The proportion of partial versus full excision biopsies has been proposed of a measure of care quality, but its usefulness has been questioned, suggesting the need for further development. 103,104 Delays in care were assessed by examining lead times between initial contact to diagnosis and treatment, to identify where care can be improved, particularly for patients with more advanced skin cancer. 81 Caution is needed, however, as lead times may also reflect the time needed to engage family in treatment planning, and to manage complex patients, factors which must be controlled for when comparing delays in care. 81,89 Surgical performance was the common focus of treatment process quality, often assessed from histopathology reports, to calculate the proportion of lesions excised, 76 and the proportion of excisions that were complete. 23 Some concerns with excision performance as a measure of quality relate to inaccuracies in GP recording of histopathological clearance, 92 whether 'near to' excised lesions were considered complete, 94 selection bias in the subset of patient data examined 17,94 and lack of longer-term follow-up of recurrence rates to definitively establish surgical quality. 92 Many articles assessing diagnostic and treatment qua lity used medical records as their primary data source. Medical records depend heavily on sound documentation-which is often lacking. 42,48,67 Incomplete records could potentially lead to underestimating GPs diagnostic accuracy, 17,67 or fail to document patient risk factors contributing to excision, 82 or misrepresent surgical adequacy, 92,94 or inaccurately depict follow-up care. 67,88 Inaccurate or incomplete documentation, and lack of standardisation in histopathological data collection and analysis systems, are major barriers to the reliability of audit and feedback. [105][106][107] Relatively few articles assessed interpersonal aspects of care. Two discussed patient-centred communication during care delivery, 58,88 while patient experience postcare was assessed in two articles through patient questionnaires. 64,88 Increasing commitment to patient-centred care suggests that facilitating shared decision-making could be explored in skin cancer care. 108,109 Structural measures of quality from the included articles related to diagnostic tools and equipment, practitioner education and training, and protocols and documentation systems (separately for diagnosis and treatment). Two of the three RCTs included in this review addressed the effectiveness of skin inspection aids and imaging systems on diagnostic accuracy. 62,64 Two articles investigated the feasibility of implementing diagnostic aids into practice, 62,63 and two looked at barriers to implementation. 61,78 A common challenge cited was that tools are usually evaluated in specialist settings rather than primary care populations 62,78,84 which have lower incidence on presentation and lower patient volumes.
Documentation systems across diagnosis and treatment included visual examination checklists, 64 dermoscopy algorithms 58 and case report forms. 63 Education and training programs were often assessed as part of interventions to improve clinical practice 62,64,65 or in reviews evaluating diagnostic accuracy. [59][60][61] Structural measures, on their own, provide limited inferences about care quality, 110 but often relate to minimum or ideal standards.
Outcome measures were also identified in the reviewed articles, including externally recorded outcomes and patient-reported measures. Externally recorded outcomes included post-treatment complications and adverse events (e.g. hospital admissions 67 ), post-treatment skin cancer recurrence, 83 and longer-term morbidity (e.g. rate of metastasis 97 ) and mortality. 68 Although outcome measures can be used to detect trends and identify outliers, 102 their validity and reliability as quality indicators is contentious due to the multitude of patient-and measurement-related confounders. 44,[110][111][112] Evaluation of commoner outcomes can be improved by controlling for population risk and other covariates 113,114 ; rarer outcomes like mortality, however, are acknowledged as insensitive measures of care quality even after adjustment except at the macro level. 115 Patient-reported outcome and experience measures are increasingly a focus of quality measurement, 116 collected prospectively in two included articles. 64,88 Patient perceptions of skin cancer treatment outcomes can substantially influence their health and quality of life, 117 but PRMs are challenging to implement in routine practice due to time and cost constraints, 96 limiting their routine deployment.
Data sources used to assess care quality must be valid and reliable, considered appropriate by clinicians and patients, and feasible to implement in practice. 40,110,118 Structure, process and outcomes of care are inherently linked, so the relationships between them must be understood for a comprehensive assessment of healthcare quality in different settings. 44,45,111 Ideally, RCTs could provide evidence that compliance with specific structure and process quality measures leads to improvements in specific outcomes. 45,110

Strengths and limitations
This scoping review cast a wide net to capture the ways in which quality has been conceptualised in primary care skin cancer management over the last decade. The thematic framework identified presents broad groupings of the structure, process and outcome quality measures proposed in primary care skin cancer management and can help to inform the development of primary care guidelines, from which indicators can be derived. This review has several limitations. Although the search strategy was designed to comprehensively capture a broad scope of quality measurement, the search terms selected may not have adequately captured literature related to key issues such as the administrative structures and organisation of services that contribute to care quality. 111 In radiation therapy for cancer, for example, facilities are regularly surveyed, within and across nations, to inform guidance on minimum or ideal resource levels. 119 In addition, restricting our database searches to articles indexed with keywords related to 'quality indicators' may have led to the exclusion of important articles on primary care skin cancer management. For example, a reviewer brought to our attention an important article 27 that addresses dermoscopy use, which was not identified through our searches or through snowballing and did not meet our inclusion criteria. It is important to note that the authors reviewed the ineligible article and concluded that had it been included it would not have altered the groupings we derived from thematic analysis of the included papers. While the weaknesses of the search strategy may detract from the richness of the data, this example suggests that the groupings derived from the included articles are robust.
As a separate limitation, we aimed to capture important quality measures suggested or proposed by each article, but it is beyond our scope to analyse in detail each individual finding as a potential indicator. It was also beyond our scope to attempt to draw conclusions about the groups or subgroups that are of greatest priority; feasibility of measurement is important to identifying indicators suitable for early adoption, but ultimately a comprehensive coverage of all the dimensions of quality is desirable. A comprehensive item-specific evidence review will be required to inform a guideline development process.

CONCLUSIONS
This scoping review has identified 13 groups of structure, process and outcome measures that have been suggested or proposed to assess quality in skin cancer management in primary care settings. This review highlights the range of areas in which relevant indicators need to be considered for development.

ACKNO WLE DGE MENTS
The authors thank Lieke van Baar, Dr. Karen Hutchinson, Mia Bierbaum and Dr Chi Yhun Lo for their contribution to title and abstract screening, and Professor Rachael Morton and Associate Professor Victoria Marr for their contribution to the study design. The authors also thank Mary Simons for her specialist guidance on database searches. Open access publishing facilitated by Macquarie University, as part of the Wiley -Macquarie University agreement via the Council of Australian University Librarians.