Enhancing ovarian cancer care: a systematic review of guideline adherence and clinical variation

Published journal articles reporting studies of any design, including literature reviews, were eligible for the study if they were available in English and published in the last 11 years (2007 to November 2018). This timeframe was chosen to ensure the inclusion of journal articles relevant to contemporary healthcare. To be eligible, studies had to focus on female adults (over the age of 18 years) who had received a diagnosis of primary ovarian, fallopian tube or peritoneal cancer. To be included, the reported outcomes had to focus on guideline adherence and its deviation, the key factors associated with variation in guideline adherence and what quality improvement approaches have been used and evidence of their effectiveness. Conference abstracts, editorials, commentaries and grey literature were excluded. Studies focussed on the centralisation of ovarian cancer care have been defined for the purpose of this review as effective care, using Wennberg’s classification of variation [7, 8]. These studies have been reported on in a previous Cochrane review and have been excluded as out of scope for this study [15].

Keywords and synonyms for the major concepts of ovarian cancer, clinical variation and guideline adherence were developed and combined to form the search strategy. Systematic searches of MEDLINE, EMBASE, SCOPUS and web of science were completed for publications between January 2007 to November 2018. Results were merged using reference management software (endnote; Thomson Reuters), with duplicates removed. The search strategy has been included in Additional file 1. The search terms used in the Medline database search were: Ovarian cancer, ovarian neoplasms, patterns of care, guideline or guideline adherence, variation of care, clinical variation, referral and consultation, referral pathway, optimal care, and framework. The terms framework and guideline were removed for search in web of science and SCOPUS due to the volume of material returned. Reference lists were also scanned to identify additional relevant studies

One author (KW) screened the titles and abstracts of the extracted papers, identifying papers that met the inclusion criteria. A random sample of the papers (10%) were screened by a second author (RH) who independently applied the inclusion criteria. The same process was followed for the full text review. Disagreements were resolved through consensus. The following data were extracted from the studies that met the inclusion criteria: author(s), date, method, data source, country, setting, sample/participants, objective, main findings and limitations.

Data were synthesised using a narrative synthesis approach [16]. This approach was chosen to allow for a text based analysis across the diverse range of research and research approaches identified in this study. The elements of narrative synthesis include developing a theory in light of the findings from the studies, developing a preliminary synthesis, exploring relationships within and between studies, and assessing the robustness of the synthesis [16].

Of the 1651 studies excluded, 185 were excluded due to wrong study design, 28 were excluded due to focussing on cancer screening and 427 were excluded as they focused on quality improvement broadly and did not specifically focus on guideline adherence and clinical variation.

Of the 32 studies included in this analysis, 19 reported retrospective population registry or local registry analysis, seven reported retrospective medical record audits, four were literature reviews, one reported a survey with clinicians and the final study included both a literature review and medical record audit. There were no qualitative studies that met the inclusion criteria. The studies were primarily from the United States (17 studies) and Europe (12 studies), with two from Canada, and one from South Korea.

Using a narrative synthesis approach to analyse the 32 papers, three broad groupings of publications were identified: 1) adherence to and deviation from guidelines (11 studies which were local, national or international) [3, 17‐26]; 2) factors impacting guideline adherence (12 studies) [27‐38] and 3) quality improvement approaches (9 studies) [39‐47].

Of the 11 studies grouped under guideline adherence and deviation, 8 utilised the NCCN ovarian cancer guidelines as the comparator guideline, with 3 using their own local or national guidelines [3, 17‐26]. These studies collectively indicated variation in adherence to guidelines for both surgical staging and receipt of chemotherapy. Bristow et al. reported that across all ovarian cancer cases in the Californian Cancer Registry, NCCN adherence was 37.2% over a 7 year period [17]. This compares to Phippen et al. who investigated NCCN guideline adherence in a low volume institute and reported that 85.4% of patients received NCCN guideline adherent surgery over an 8 year period [23]. One of the key factors distinguishing these studies was that in the Phippen et al. study all surgery was completed by gynaecological oncologists, with surgeon type not reported in the Bristow et al. study [17, 23]. Erickson et al. found a higher rate of guideline adherent care, at 78.5%, in their single institution only study [19]. This is in contrast to the population study by Hodeib et al. using data from the Californian Cancer Registry, finding guideline adherence of 24% [21]. See Table 1 for characteristics of the studies included under guideline adherence and deviation.

Individual hospital and/or surgeon volume and their link to improved outcomes has been identified for surgery that is technically complex [48]. Hospital and/or surgeon volume emerged as the primary factor associated with the degree of guideline adherence identified in 11 of the included studies in this review that in turn was linked to ovarian cancer outcomes. Eight of the 11 studies identified higher volumes with improved outcomes at a national population level, with three studies focussing on higher volume leading to improved outcomes at a state or specific geographical level [27‐32, 34‐36, 38]. Survival at both 5-years and 4-years was found to be better for women treated in higher volume hospitals [29, 30, 32, 38]. The Marth et al. and Bristow et al. studies found hospital volume was an independent predictor of survival [29, 36]. Hospital volume was also found to be a statistically significant and independent predictor of optimal cytoreductive surgery, i.e. guideline adherent care [28, 29]. Shakeel et al. and Wright et al. reported that higher volume hospitals had lower in-hospital mortality and morbidity [33, 34]. The study by Uppal et al. investigated the use of 30-day readmission following cytoreductive surgery as a measure of quality [49]. This study found that although higher volume hospitals had higher rates of 30-day readmission, they also had higher rates of guideline adherent care and an improved 5-year survival when compared to lower volume hospitals with lower 30-day readmission rates [37]. This supports their proposal that 30-day readmission following ovarian cancer surgery is not a robust measure of quality care. Overall the studies found that improvement in guideline adherence was associated with improved outcomes in hospitals where there was high volume ovarian cancer care at a hospital and/or surgeon level. See Table 1 for characteristics of the studies included under factors impacting guideline adherence.

There were nine studies identified investigating quality improvement approaches that attempted to detect, respond to or address clinical variations. Five of the studies focussed on the development of quality indicators for guideline adherent care as a way to identify variation in clinical practice and improve outcomes [40, 44‐47]. Three studies reported on the impact of quality improvement programs on improving outcomes in ovarian cancer through increasing adherence to clinical guidelines, resulting in a reduction of unwarranted clinical variation [39, 41, 42]. The final study reported on one hospital’s compliance with Society of Gynecologic Oncology quality indicators [43]. Five studies reporting on the development of quality indicators all had similarities in the indicators developed, with a focus on rates of complete surgical resection [40, 44‐47]. Three of these studies focussed further on the type of surgeon and their surgical volume [40, 45, 47]. The study by Phillips et al. focussed on the importance of identifying the denominator in advanced ovarian cancer studies and quality indicators, to ensure outcomes are able to be appropriately interpreted and the ‘denominator effect’ is understood [44]. Phillips et al. defined the denominator in their study as the total number of advanced ovarian cancer cases presenting or referred to the cancer centre [44]. This then allowed them to understand the proportion of patients who received treatment and the cancer centres overall survival rates for women with advanced ovarian cancer [44]. The study by Aletti et al. found an increase in complete cytoreductive surgery from 31 to 43% following the implementation of a quality improvement program [39]. The quality improvement program implemented included comparison of hospital results with the latest evidence, confidential surgeon benchmarking, educational sessions for trainees on new and complex surgical techniques and improving the availability of experienced staff members to assist with complex surgical cases [39]. The study by Kommoss et al. found women with stage 1A-IIIA ovarian cancer had an improvement of 27 to 88.5% in complete cytoreductive surgery following the implementation of a quality improvement program [42]. Harter et al. further reported on the same quality improvement program for women with later stage disease, finding complete cytoreductive surgery rates for women with stage IIB-IV ovarian cancer improved from 33 to 62%, with overall survival improving from 26 months to 45 months following the implementation of the quality improvement program [41]. The quality improvement program reported on in the Kommoss et al. and Harter et al. studies focussed on optimal surgical staging, receipt of platinum based chemotherapy, surgeon volume and experience, as well as the implementation of dedicated ovarian cancer surgical teams, benchmarking of morbidity and outcomes, and improvement of interdisciplinary assessment and management [41, 42]. There are commonalities between the three studies that reported on outcomes of the implementation of quality improvement programs, including adherence to guideline based care, improvement in ongoing surgical education and surgeon benchmarking [40‐42]. The study by Liang et al. found high clinical variation in the indicators of complete surgical staging and the administration of intraperitoneal chemotherapy within 42 days of optimal cytoreductive surgery [43]. See Table 1 for characteristics of the studies included under quality improvement approaches.

Given the relationship between hospital and surgical volume, guideline adherence and care outcomes, our findings suggest that there is value in programs to promote guideline adherent care in hospitals with high surgical volume, at both a hospital and surgeon level [30, 33]. Development and implementation of quality indicators and quality improvement programs to improve ovarian cancer outcomes have been used internationally in an attempt to reduce unwarranted clinical variation in ovarian cancer care [39‐43, 45‐47]. A core strategy associated with such programs is the centralisation of care to hospitals that have specialist gynaecological oncology centres and complete a high volume of ovarian cancer surgery at both the hospital and surgeon annually [51‐54]. Evidence suggests that centralising ovarian cancer care into specialist hospitals improves access to gynaecological oncologists, multidisciplinary cancer care teams and higher surgical volume hospitals and surgeons may lead to improved survival [15].

The development of gynaecological oncology as a surgical sub-speciality in Australia, the UK and North America has led to increasing specialist care in the treatment of gynaecological cancer. In a study identifying practice guideline adherence in Ontario, Canada, 81% of gynaecological oncologists reported completing surgical staging adherent to guidelines, with only 41.5% of non-oncologists completing guideline adherent surgical staging [3]. A study by Chan et al. found treatment by a gynaecological oncologist led to more women in California to undergo primary staging surgery (91.9% vs. 69.1%) and chemotherapy (90% vs. 70.1%) than those treated by a non-oncologist [55].

An important focus for future research is to establish evidence of the effectiveness of the implementation of quality improvement programs, reporting on survival over time. Evidence of the relationship between quality improvement programs and survival is important for assessing the impact of implementing quality indicators through quality improvement programs as a way to decrease unwarranted clinical variation and improve outcomes for women living with ovarian cancer. Analysis of the implementation of quality indicator and quality improvement programs across regional or national jurisdictions that aim to improve outcomes in ovarian cancer beyond a single institution level would be valuable.

The included studies and review process were subject to some limitations. Whilst hospital and surgeon volume were identified as important, a key challenge for interpreting the results of these studies was the range of case volume for hospitals, with each study defining volume in different ways. High hospital volume ranged from > 20 to > 30 cases/yr. [29‐32, 36]. High surgeon volume was more consistent, being identified as ≥10 cases/yr. in the two studies by Bristow et al. that investigated surgeon level volume [27, 28]. For health systems undertaking centralisation of ovarian cancer surgery this may lead to challenges in identifying which hospitals should be completing ovarian cancer surgery. Reliance on administrative datasets created limitations such as selection bias, as most registries did not cover the entire population, with a lack of information on stage of disease, extent of cytoreductive surgery and surgeon specialist type. Two examples are the study by Erickson et al. as a single institute study finding a high rate of guideline adherent care, at 78.5%, and the study by Hodeib et al. which used population data and found a low rate of guideline adherent care at 24% [19, 21]. A limitation in study design, using single institution versus population data, may in some way explain the vast difference in guideline adherence between the two studies [19, 21]. The review process may have omitted relevant material by including only studies published in English, the exclusion of studies allocated to broader quality of care issues, as there are many factors associated with clinical variation, guideline adherence being just one, and the date range of the past 11 years.