Introduction

Europe is one of the largest markets for the fast-growing sector of medical devices (MDs) and diagnostic procedures, which encompass a broad and heterogeneous range of technologies. Due to the rising costs associated with introducing of new MDs and procedures into the healthcare system, payers have started to pay more attention to the effectiveness and financial implications of such new technologies. In this context, health technology assessment (HTA) has gained increasing recognition at the European level as a decision support tool [1].

No other medical technology has been evaluated as frequently by HTA institutions in European countries as “positron emission tomography (PET)”, or rather “positron emission tomography/computed tomography (PET/PET–CT)” [2, 3]. In its 2015 assessment report [4] on PET/PET–CT, the Ludwig Boltzmann Institute for Health Technology Assessment (LBI-HTA) aimed to capture the evidence base on oncological indications for a needs-based planning of PET/PET–CT facilities in Austria. It identified around 160 assessments on PET or PET–CT in a 10-year period (2004–2014), of which the first HTA were already published in 1995. This expresses uncertainty about the value of PET diagnostics in patient care. Notably, despite ongoing and controversial discussions about the patient-relevant benefits of PET/PET–CT, these technologies have rapidly been adopted both in Europe and abroad [2]. Additionally, the majority of those assessments is not accompanied by decisions regarding the reimbursement or planning of PET/PET–CT indications or devices. However, several approaches exist to deal with (to varying degrees) uncertain evidence or contradictory recommendations around reimbursement and capacity planning in Europe and globally [4]. In practice, capacity planning employs a mix of different methods and methodological approaches such as comparative approaches (e.g., benchmarking), analytical (e.g., health care needs-assessment) and/or reactive approaches (e.g., waiting lists). To provide volume forecasts for facility planning and workload vis-à-vis PET/PET–CT, the number of patients with (sub-)indications is important as is the estimated number of patients eligible for therapy monitoring and radiotherapy dose planning. In addition, some special features of PET/PET–CT (e.g., required radionuclides/tracers) have to be taken into account. Yet the planning of capacities based on evidence-based indications is rarely seen—neither in the literature nor in practice.

On the initiative of the German Society for Nuclear Medicine (DGN), the LBI-HTA and the Department of Health Care Management at the Berlin University of Technology collaborated to (1) update the LBI-HTA 2015 report to identify indications recommended by evidence [HTA, systematic reviews/meta-analyses (SR/MA), evidence-based guidelines (EBG)] for PET/PET–CT scans supporting facility planning in Germany and Austria and (2) describe a pilot study which applied those recommendations identified in the evidence review from step (1) to an investment decision in the Austrian region of Carinthia. The subsequent, updated, HTA report extended the scope of the 2015 report by including indications for neurology and inflammatory disorders in addition to oncological indications.

The following article summarizes both the results of the updated PET–HTA report, which has already been published (in German [5]), and the pilot study in Austria testing the practicalities of the recommendations.

Materials and methods

(1) Evidence-based indications and recommendations

Sources of information, search strategy, and study selection

A systematic literature search was carried out in July 2017 to identify HTA reports, SR/MA, and EBG to be included in the study. The databases MEDLINE, EMBASE, PubMed, and Cochrane Library were searched using a search strategy retrieved from a published report by the German Institute for Quality and Efficiency in Health Care (IQWiG), which was then updated and adapted [6]. A 5-year window (2012–2017) for EBG and SR/MA was chosen [7]. HTA for the newly included indications (neurology and inflammatory disorders) were limited to the last 10 years (2007–2017), while the most recent HTA [8–10] included in the PET–HTA report [4] were used for oncological indications, though oncological guidelines already included in the PET–HTA were updated if newer ones were available. Details about the searches are found in the Appendix A of the report [5] and a translated version is found in the Online Resource 1 of this article.

The literature was screened in a two-stage process (First: title/abstract; second: full text) by two researchers of the author group. Inclusion and exclusion criteria were determined based on the criteria from the PET–HTA report 2015 [4] for consistency. English- and German-language HTA, SR/MA, and EBG on the benefits of PET/PET–CT were included, if they discussed (1) patient-relevant benefits; (2) diagnostic accuracy/quality or change in patient management; (3) if they focused on one of the three indications/treatment areas (oncological and neurological indications, or inflammatory diseases); (4) were developed with an evidence-based methodology (e.g., a description of a literature search referenced more than two databases, inclusion/exclusion criteria and quality assessment, and double-check principle); and (5) were freely available. Differences in the assessment of literature were resolved through discussion and consensus building. A detailed overview of the criteria for inclusion and exclusion can be found in the Online Resource 2 of this article.

In addition to the systematic search, a supplementary, comprehensive manual search for evidence-based recommendations on the use of PET/PET–CT was conducted. The selection of databases or websites [e.g., Cancer Care Ontario (CCO) and National Comprehensive Cancer Network (NCCN)] was based on the PET–HTA report 2015 [4] and complemented by relevant sources for the two new indications. The search could not be carried out in a systematic manner due to variation in search interfaces, but the search term PET was a common factor. Up-to-date recommendations of the following national and supranational societies of nuclear medicine were included: Austrian Society of Nuclear Medicine and Molecular Imaging (ÖGN), German Society for Nuclear Medicine (DGN), Swiss Society for Nuclear Medicine (SGNM), European Association of Nuclear Medicine (EANM), Society of Nuclear Medicine and Molecular Imaging (SNMMI), and Joint Collaboration of EANM and SNMMI. Databases (e.g., Choosing Wisely, NICE Do-not-Do database) that explicitly identify “inappropriate” (not-to-do) services were also included as a source of information. Screening of documents from websites or professional societies was carried out by one person, with the extraction performed using the double-check principle. During the process of updating the HTA report, information sources were again scanned between January and February 2018.

Data extraction

Extraction tables were compiled separately for the three indications, using the double-check principle. In a first summary table, study characteristics were extracted separately, because they differ among HTA (e.g., reported endpoints and number of included studies), SR/MA (e.g., reference standard), and EBG (e.g., funding/sponsoring and strength of the evidence). Evidence derived from HTA and SR/MA was thus presented in the further tables including the evidence base and verbatim conclusions. The recommendations from EBG and websites and databases of national and supranational societies were first extracted in tabular form and assessed along two dimensions: (1) appropriate use criteria and (2) inappropriate use criteria (e.g., suspension of decisions, recommendations based on insufficient evidence, and disinvestment recommendations), including also the corresponding strength of the evidence. Recommendations of a certain level/strength of evidence (see Box 1) were then considered and clustered according to type and content of recommendations for further analysis into their similarities and differences.

(2) Pilot study: evidence-based PET/PET–CT planning

A study was carried out to guide the investment decision for an additional PET–CT scanner in the region of Carinthia in Austria (561,000 inhabitants). First, a needs-assessment based on a data analysis of PET scans was conducted. Diagnoses were coded according to the International Statistical Classification of Diseases and Related Health Problems 10th revision (ICD-10, 2016 [14]). All PET scans performed from January to September 2017 were retrieved and clustered according to disease diagnosis (e.g., tumour type: C.00; malignant neoplasm of lip) and, if available, to subgroups of indications (e.g., C00.1 external lower lip). The information was then matched to identify recommendations for evidence-based indications or non-indications [see step (1)]. Indications with contradictory recommendations were—due to lack of detailed clinical information—ascribed to the category of dissenting or unclear evidence from HTA and/or guidelines. The categorization as well as the matching with evidence-based indications was performed by two researchers.

Results

(1) Evidence-based indications and recommendations

Results of the evidence selection process

The systematic search yielded 10,873 references (Fig. 1). After the removal of duplicates (2729 references) and the—ex post—introduction of a time limit (due to an enormous amount of materials), 4299 references were ultimately excluded and 3845 references remained for the two-step screening. Of these, 3412 references were excluded according to the inclusion and exclusion criteria, leaving 433 full texts, 43 of which were finally included (10 HTA reports, 20 SR/MA, and 13 EBG). An overview of the excluded references (n = 390) with reasons can be found in the Appendix D of the updated HTA report [5]. The screening of the collected alerts of the database searches (update: from July to December 2017) resulted in three more HTA, one EBG, and three SR/MA. As such, a total of 47 references were included.

Fig. 1
figure 1

PRISMA flowchart: study selection process of the systematic search [5]. *Due to the large amount of records identified, a time limit was set up for the title/abstract screening of SR/MA. **Online Appendix D [5] contains a list of excluded articles with reasons

Full size image

The updated search results in guideline databases and websites and databases of relevant national and supranational societies regarding recommendations (see tables in report [5]) was compared with the results of the systematic research and the PET–HTA Report 2015, resulting in additional “appropriate use” recommendations from seven professional societies: DGN and the joint collaboration between EANM and SNMMI, RCP/RCR, ACR Appropriateness Criteria®, CAR Referral Guidelines, and two oncology expert networks for guideline development, CCO and NCCN. It also resulted in concrete “not-to-use” recommendations from four societies or databases: ACR, CAR, and the two disinvestment databases Choosing Wisely and NICE Do-not-Do database. Thus, a total of ten HTA, and 234 positive and negative recommendations from professional societies and databases were included, supplemented by the statements from 23 SR/MA.

Evidence on oncological indications

Study characteristics

The PET–HTA report (2015) included 35 HTA assessments on 20 oncological indications. In addition, the report update revealed two novel HTA on bronchial carcinoma from the US Agency for Healthcare Research and Quality (AHRQ) [15] and the HTA-Center for the Västra Götaland Region, Sweden [16] as well as one HTA each on mamma carcinoma, compiled also by AHRQ [15], on penile/testicular carcinoma and on bladder/renal cancer, the latter two compiled by the Scottish Health Technologies Group within Healthcare Improvement Scotland (SHTG/HIS [17–20]). For further oncological indications considered in the PET–HTA report 2015, no more recent HTA was found. Reported endpoints in the included studies used for the HTA reports refer to diagnostic accuracy (mainly tumour grading and change in management).

Recommendations as well as non-recommendations (n = 188) from professional societies (see Box 1 for abbreviations) were retrieved and extracted from AANS/CNS [21], ACCP [22], BSH [23], CCO [24–26], DGN [27], IMWG [28], and RCR/RCP [29]. Three common guidelines, one from SNMMI/EANM and others (ACNM, ACPM, ASCO, CANM, and SPR) [30], one from RANO/EANO [31], and one from EANM/SNMMI [32], were also considered. One guideline (Ryken et al. [33]) has been elaborated by several universities.

Three guidelines (CCO [24], RCR/RCP [29], SNMMI/EANM/others [30]) provided recommendations for tumours generally as well as for non-oncological indications. Another three guidelines dealt with tumours of the head (DGN [27], Ryken [33], RANO/EANO [31]); two focused on myeloma (BSH [23], IMWG [28]); and one guideline addressed prostate cancer (EANM/SNMMI [32]), anal canal carcinoma (CCO [25]), lung carcinoma (ACCP [22]), paraneoplastic syndrome (PNS) (CCO [26]), and pituitary adenoma (AANS/CNS [21]), respectively. The recommendations of ACR [34], CAR [35] and NCCN [36] as well as the explicit non-recommendations of choosing Wisely [37] and NICE Do-not-Do [38] were also considered.

Moreover, 12 SR/MA for seven oncological sub-indications were included [39,40,41,42,43,44,45,46,47,48,49,50]; of those, two supplemented an HTA report [49, 50] and all considered diagnostic accuracy as an endpoint, especially for primary diagnosis and staging for pre-treatment/treatment planning. The systematic reviews consisted primarily of retrospective studies.

Quality assessment

The quality of the HTA was overall very good, with only a few questions not evaluated with “yes”. The assessment of the quality of the guidelines varied. In domain 1: “scope and purpose” and domain 4: “clarity of presentation”, the guidelines received high ratings and there were fewer variations among the guidelines. Domain 3 (“rigour of development”), which considers the accuracy of guideline development, ranged from 11% (DGN [27]) to 93% (ACCP [22]), showing the largest differences among the guidelines. In the overall rating, the ACCP guideline [22] was given a score of 5.67 (max. 7) and also ranked highest among two domains 3 and 4. The RCR/RCP guideline [29] attained last place with a mean score of 2.67.

The quality of the SR/MA in terms of overall qualitative assessment (no overall score) revealed that 10 out of 12 were “critically low”. Hence, the SR/MA is used only as supplementary information.

Overall recommendations

Table 2 provides an overview of the overall recommendations on oncological indications (see also Table 1 for the explanation of colors and categories).

Table 2 PET/PET–CT indications (oncology): overall recommendations
Full size table

There is a (relative) consensus that there is sufficient evidence for sub-indications in eight indications in favour of PET or PET–CT examinations (in Table 2 highlighted green). The first six were already determined in the 2015 report—(1) bronchial carcinoma (update: mainly pre-treatment, contradictory in re-staging and response control and in therapy monitoring), (2) colon carcinoma, (3) malignant lymphoma, (4) malignant melanoma (update: contradictory for diagnosis of recurrence), (5) mamma carcinoma (treatment response, for diagnosis of recurrence), and (6) head–neck tumours (in 2015 report: CUP, thyroid carcinoma; update: mainly for diagnosis of recurrence)—while two new treatment areas were added by the update: (7) myeloma and (8) neuroendocrine tumours.

There is a (relative) consensus that, in eight other indication areas, there was too little evidence in favour of PET examination (individual decisions possible) (in Table 2 highlighted red): (1) bladder carcinoma, (2) hepatic cancer, (3) cervical carcinoma, (4) gastric cancer, (5) ovarian and (6) uterus carcinoma, (7) prostate cancer, and (8) paraneoplastic neurological syndrome.

For a further set of eight indications, there is contradictory and equivocal evidence and recommendations were developed along with reservations (in Table 2 highlighted yellow or blue): (1) anal canal, (2) brain (especially glioma), (3) testicular, (4) renal, (5) penile carcinoma, (6) esophagus cancer (except re-staging), and (7) pancreatic carcinoma as well as (8) bone and soft-tissue tumour (+ gastrointestinal stromal tumour).

Evidence on neurological indications

Study characteristics

Two HTA reports were identified about neurological indications for the use of PET—one from the Australian Medical Services Advisory Committee (MSAC) [51] looking into Alzheimer’s dementia (AD) and one compiled by the Canadian Agency for Drugs and Technologies in Health (CADTH) on epilepsy [52]. Reported endpoints in the included studies of these HTA reports refer mainly to diagnostic accuracy; no study could be identified that investigates diagnostic effectiveness and safety.

Recommendations and non-recommendations (n = 28) from professional societies (see Box 1 for abbreviations) were retrieved and extracted from CCO [24, 53], EFNS [54], RCR/RCP [29], SNMMI/AA [55], and a common guideline from EANM/SNMMI [56]. RCR/RCP [29] provides recommendations for several non- and oncological indications. Two further guidelines ([54, 55]) consider Alzheimer’s dementia/dementia (ADD) or AD only, while another guideline (EANM/SNMMI [56]) presents general recommendations on brain-related disorders (refer to SNMMI/AA [55]) CCO [24, 53] (two guidelines) which considers epilepsy. In total, three SR/MA were included (for AD) [57,58,59]. All consider diagnostic accuracy in terms of primary diagnosis as the endpoint. Primarily, prospective studies were included in the SR/MA.

Quality assessment

The quality of the HTA was rated as very good, with only one question not judged as “yes”. The assessment of the quality of the guidelines showed that domains 1 (“scope and purpose”) and 4 (“clarity of presentation”) received very high ratings and are quite consistent. However, in terms of editorial independence, applicability, and accuracy of the guideline development, big differences were observed. In the overall assessment, SNMMI/AA [55] ranks first; RCR/RCP is in last place [29]. All the SR/MA rated “critically low” in the overall quality assessment (no comprehensive score). As such, the SR/MA are to be used as supplementary information for the overall recommendations.

Overall recommendations

For recommendations of neurological indications, evidence identified and presented only two sub-indications: Alzheimer’s dementia/dementia and epilepsy (see Table 3 for details).

Table 3 PET/PET–CT indications (neurology): overall recommendations
Full size table

There is a (relative) agreement that there is not sufficient evidence in favour of a PET/PET–CT for either of these two sub-indications (in Table 3 highlighted red), though professional societies uniformly/consistently name specific cases of AD or specific patient characteristics that speak for/against the use of PET, depending mainly on the respective tracer (amyloid vs. FDG). These recommendations, however, are only based on one source [55, Update: 60] and the authors themselves acknowledge the limitations, stating that, “At the time of this review, experience with clinical amyloid PET imaging [was] limited. Most published studies to date … [were] designed to validate this technology and understand disease mechanisms rather than to evaluate applications in clinical practice. As a result, published data are available primarily from highly selected populations with prototypical findings rather than from patients with comorbidities, complex histories, and atypical features often seen in clinical practice (e7)” [55].

Equivocal evidence (e.g., contradictory between HTA and EBG in Table 3, highlighted yellow or blue) was also found for the application of PET/PET–CT in patients having epilepsy (again, only in certain cases, in specialized epilepsy centres), though there is (some) consensus among the professional societies.

Evidence on inflammatory disorders

Study characteristics

In total, three HTA could be included about inflammatory disorder indications. One focused on infections in general from CADTH [61], dating back to 2008, and two other reports (each consists of a scoping report and an advice statement) from SHTG/HIS (2013) focus on pyrexia of unknown origin and sarcoidosis [62–65]. Reported endpoints in the studies within these HTA reports refer to diagnostic accuracy (mainly for primary diagnosis).

Recommendations and non-recommendations (n = 18) from professional societies (see Box 1 for abbreviations) were retrieved and extracted from ECCO/ESGAR [66], a common guideline of EANM/SNMMI [67] and RCR/RCP [29] who gave recommendations for several non-oncological indications. EANM/SNMMI [67] gave general recommendations on inflammatory disorders and ECCO/ESGAR [66] considers inflammatory bowel disease alone. Explicit non-recommendations by Choosing Wisely [37] or NICE Do-not-Do [38] were not identified.

In total, eight SR/MA were included for five indications [68,69,70,71,72,73,74,75]. All but two reviews [68, 69] include a meta-analysis and consider diagnostic accuracy as the primary endpoint.

Quality assessment

The quality of the three HTA was rated “very good”, with only a few questions that could not be judged “yes”. And the quality of the three guidelines revealed that domain 3 [on the rigour of the guideline development (including evidence clearly assigned to recommendations)] ranges widely, from 16% (RCR/RCP [29]) to 55% (ECCO/ESGAR [66]). Notably, all guidelines were rated 0% with regard to editorial independence (domain 6). The RCR/RCP [29] guideline scored poorest on in all areas compared to the other two. In the overall rating, ECCO/ESGAR [66] ranked first of the guidelines and all SR/MA were rated as “critically low”.

Overall recommendations

Table 4 shows the overall recommendations regarding the appropriate or inappropriate use of PET/PET–CT for inflammatory disorder indications. There is (relative) consensus around sufficient evidence in favour of PET or PET–CT for infections of the vertebral column/spondylodiscitis. For the following four sub-indications, however, there is contradictory or inconclusive evidence: periprosthetic joint infection, osteomyelitis, sarcoidosis, and fever of unknown origin.

Table 4 PET/PET–CT indications (inflammatory indications): overall recommendations
Full size table

(2) Pilot study: evidence-based PET/PET–CT planning

In a second step, the recommendations derived from the evidence were applied (matched) to the hospital data on PET scans in one hospital (Carinthia, Austria) to gain an understanding of the PET utilization and capacity needs. Between January and September 2017, 1762 PET scans were conducted at the Clinicum Klagenfurt/Carinthia (KABEG Management: Needs-assessment of PET–CT at the Clinicum Klagenfurt, June 2018, unpublished). Of those, 1370 (77.8%) could be assigned to the three categories (see Tables 1–4) as recommended by evidence-based indications (963, 54.7%), not recommended (311.8%) and contradictory recommendations (376, 21.3%). The other 392 (of 1762, 22.2%) could not be allocated to any of the three treatment areas due to missing information of the diagnostic code or lack of information from recommendations in HTA or clinical guidelines.

A systematic analysis of all 1762 PET scans was not possible for two reasons: first, all ambulatory patients who underwent a PET scan were coded with a non-specific diagnostic code (such as ‘other’ investigation), so that the underlying reasons for PET diagnosis could not be identified and, second, no diagnostic code was assigned to in-patients transferred for the PET scan from the other Carinthian hospitals. Of the 963 PET scans that were possible to match to evidence-based recommendations (see Table 5), four ICD-10 tumour categories alone accounted for 79%: C30–C39: malignant neoplasms, respiratory system, and intrathoracic organs (273, 28%), C43–C44: malignant neoplasms, skin, (224, 23%), C81–C96: malignant neoplasms, stated or presumed to be primary, of lymphoid, haematopoietic and related tissue (177, 18%), and D37–D48: neoplasms of uncertain or unknown behaviour (99, 10%).

Table 5 PET scans and corresponding ICD-10 category matched to evidence-based recommendations (n = 994), Clinicum Klagenfurt/Carinthia (KABEG Management: Needs-assessment of PET–CT at the Clinicum Klagenfurt, June 2018, unpublished)
Full size table

The majority (22, 70%) of the 31 PET scans not recommended within evidence-based recommendations (see Table 5) were in the following three ICD-10 indications: C64–C68: Malignant neoplasms, urinary organs (10, 32%), C15–C26: malignant neoplasms digestive organs (6, 19%), and C50–C58: malignant neoplasms, breast and female genital organs, respectively (6, 19%).

Discussion

Summary of findings: (1) evidence-based indications and recommendations

Evidence assessments on PET/PET–CT have been increasingly published over the past two decades in many countries and languages. Taking a 2004–2014 perspective, the PET–HTA 2015 report identified 155 HTA reports to evaluate PET or PET–CT; even with a timeframe of 2008–2014, there were still 82 HTA available. Thirty-five of these were included and extracted for their statements on the use of PET/PET–CT for oncological indications. Five more recent HTA on oncological (sub-)indications were identified during the HTA report update. For the new indications—for neurological and inflammatory disorders—only five HTA could be considered. Recommendations of professional societies (e.g., guidelines) were also updated and explored for the three indications. A total of 234 recommendations were, therefore, included in the updated report. In addition, a total of 23 SR/MA provided supplementary information. Fifteen looked only at oncological indications. Cost-effectiveness of PETs was not an inclusion criterion and was, therefore, not explicitly considered; however, while almost no evidence exists, information was extracted from the reports when there was mention of cost-effectiveness.

In a comparison of HTA results, the tabulated recommendations of professional societies and SR/MA for oncological indications reveal that there is general agreement on “significant” indications (e.g., bronchial carcinoma, and head and neck tumours) for the use of PET/PET–CT. However, there are significant differences in the level of detail of sub-indications and in the approach for when a scan should not be performed. There is also significant variance in the reliance on the access of graded pre-diagnostics.

Between the current study and the 2015 PET–HTA report, several changes in the categorization of oncology indications and in level of contradictory among the evidence sources have occurred (below is an overview; see Tables 1, 2, 3, 4):

  • Mamma carcinoma (now recommended for certain sub-indications such as tumour response assessment for metastatic mamma carcinoma (mainly bone metastases), before: not recommended)

  • Cervix carcinoma (not recommended, before contradictory)

  • Testicular, renal, and penis carcinoma (contradictory, before not recommended)

  • New (not included in the PET–HTA report 2015): myeloma (recommended), neuroendocrine tumours not recommended), anal canal cancer (contradictory)

In the case of neurological indications and inflammatory disorders, there is comparatively more discrepancy in the recommendations of professional societies and between the HTA and professional societies. There is a notable, consistent emphasis on the weak and insufficient evidence base. The SR/MA included in our report demonstrate this: all were considered “critically low” in the quality assessment.

Summary of findings: (2) pilot study on evidence-based PET/PET–CT planning

Based on evidence-based recommended and not recommended indications formulated in step (1) of this study, a first-ever pilot of planning PET capacities was conducted and, as such, has no comparator.

A data-driven needs-assessment was initially conducted to provide a basis for an investment decision around the need for an additional PET–CT scanner in Carinthia, Austria. First, diagnoses were coded according to ICD-10 classifications. Then, the 1762 PET scans performed between January and September 2017 were matched to relevant recommendations for evidence-based indications/non-indications, resulting in 77.8% being assigned to either recommended (54.7%), not recommended (1.8%) or contradictory (21.3%) evidence-based indications. 22.2% could not be ascribed due to missing information of the diagnostic code or lack of information from recommendations in HTA or clinical guidelines. Based on the data analysis and an additional assessment of the utilization as well as occupancy rates, it was decided against the investment in a second PET scanner in Carinthia.

Limitations

The main strength of the presented work is the rigorous, systematic approach applied in all steps (e.g., double-check principle and quality assessment), complemented a comprehensive, targeted manual search. Nevertheless, the methodical process has some limitations. The systematic synthesis of available evidence, which mainly relies on already aggregated evidence, may result in a certain loss of detailed information. In addition, the primary outcome of interest in the data extraction was the conclusions as articulated by the respective authors, which are different in terms of formulation and meaning and, therefore, difficult to compare. As a result, a trend can be seen in the statements.

Moreover, the multitude of recommendations from professional organizations necessarily led to a capping of the number of sources, meaning that some were not included and important contextual insight may have been lost. For example, the S3 guidelines from the Association of Scientific Medical Societies e.V. (AWMF) in Germany and the guidelines from the Scottish Intercollegiate Guidelines Network (SIGN) were not included as they do not have a specific PET/PET–CT focus. However, they both provide important recommendations as they consider not only “study evidence”, but also detailed aspects of the diagnostic-therapeutic chain and differentiate more comprehensively among tumour entities. That some guidelines were a priori omitted may lead to a potential bias (of the overall picture). Similarly, the heterogeneous quality of the included guidelines may also result in a skewed overall picture of existent recommendations. The summary tables make this transparent by assigning respective recommendations to the corresponding guidelines, referencing, as well, the strength of evidence. Finally, in Tables 2, 3, and 4 of the overall recommendations, a differentiated consideration of tumour entities is lost. Nevertheless, they provide an overall picture that can be useful (see pilot).

Regarding the 22.2% of PET scans in Carinthia, Austria, that could not be assigned to a recommendation in the pilot study, one reason could be the very specific indications found in the evidence or a too detailed coding. In general, while the high number of PET scans that could not be ascribed to any category could be seen as limiting the explanatory/instructive power of the pilot to use evidence-based indication lists as basis for a needs-assessment and investment decisions, it also shows clearly that there is a need for improvement of coding indications within the context of planning of medical services. For the final investment decision, further information on timing and frequency of PET utilization and occupancy rate over full work days and weeks are needed.

Conclusion

Overall, this study has resulted in more detailed information and specifications around PET/PET–CT indications as compared to the 2015 PET–HTA report. This update together with the aggregate list of overall recommendations for indications as well as the explicit non-recommendations from the 2015 PET–HTA report can serve as needs-based and evidence-based decision support for PET/PET–CT service provision in hospitals as evidenced by the pilot study in Austria. A better coding of PET utilization is needed for planning. Furthermore, additional information such as timing, frequency of PET utilization, as well as occupancy rate over full work days and weeks, is also necessary to guide an investment decision.