This paper’s discussion will firstly focus on its methods followed by the results’ discussion. To finalize we will briefly discuss its contribution to science and healthcare, as well as important underlying questions about using ACSCs as an indicator.
Regarding methods we have developed the list for the Portuguese context using the Delphi Panel method. All lists of ACSC have been developed using experts opinion, being the Delphi panel method the most used.4,8,10 The reason for this is the necessary evaluation of comprehensive and complex disease mechanism and diagnostic/treatment pathways in order to determine if a condition can be avoided and/or treated in the ambulatory setting. No other valid method is described. The experts’ selection process is extremely important for the panels’ conclusions. Therefore, we intentionally mixed GPs with IM since their perspectives are complementary. While the first are responsible for most ambulatory care in Portugal, therefore understanding the resources available in the ambulatory setting, the latter are the most responsible for hospital admission decisions and have the end-of-the-line perspective on the pathways that lead to an ACSC hospitalization. To ensure the experts quality we defined clear selection criteria and arranged for a peer-selection by the most relevant Portuguese scientific societies in each specialty. The difference in the invitation acceptance rate between GPs (50.9%) and IMs (17.2%) is very likely related to the higher familiarity with the concept of ACSC by GPs, once these are used as a performance indicator of primary care by local, regional and national authorities. The final mix of 86% GPs was higher than initially designed. However, the IM were fairly active during the several rounds and positively contaminated the discussion on several items.
In terms of results the experts proposed fourteen new ACSCs having four conditions achieved consensus, namely uterine cervical cancer, colorectal cancer, thromboembolic venous disease and voluntary termination of pregnancy. For both type of cancers, the experts referred to the screening undertaken in primary care as the technology that might avoid the necessity of hospitalization. The reasoning is that early detected conditions might obliviate the need for more complex interventions that require hospital admission. This does not, however, mean that all cancers and therefore all admissions for cancer are avoidable. Regarding thromboembolic venous disease, experts identified the early diagnosis and treatment as capable of avoiding hospitalizations. Finally, voluntary termination of pregnancy was considered avoidable if effective and timely family planning is accessible. All of these conditions verified the relevance for Public Health criterion proposed by Solberg and Weissman (hospitalization rate higher than 1/10.000 hab) being colorectal cancer, in fact, the fifth highest rate of ACSC hospitalization in Portugal in 2017. Although these criteria have been widely accepted, we do not agree that Public Health relevance should be strictly reduced to a hospitalization rate. For this reason, we propose two lists of ACSC, the core list including only the conditions that verify the hospitalizations rate criterion, and the extend list where all conditions considered ACSC by experts are included.
It is also noteworthy that other four conditions strongly related to behavioural and lifestyle decisions were newly proposed and almost reached consensus, namely alcoholic liver disease, hepatitis C, HIV infection and lung cancer. The health promotion capacity of primary care was identified as the technology that might prevent the onset of these diseases and therefore the need for hospitalization. It is also relevant to discuss that the recent pharmacological innovation for the treatment of HIV and hepatitis C viruses infection might render the need for hospitalization growingly residual.
Another important discussion topic is the consensus in not considering appendicitis with complication an ACSC and the lack of consensus in further nine previously considered conditions. This fact elicits the discussion on the context specificities to which is important to adapt the ACSC lists. This may also reflect the limitation of using experts’ opinion to define ACSCs. In summary, the influx and efflux of conditions considered ACSC reflects the importance of regularly updating the lists. Nearly all the previous lists used in the comparison in Table 4 were developed more than 10 years ago. The constant development of knowledge, technology and healthcare design demands that the ACSC lists are updated more regularly.
The rates of hospitalization in this paper are in line with previous studies regarding the most frequent ACSCs in Portugal being pneumonia, COPD, cardiovascular diseases, urinary tract infection and diabetes. Using the developed list, we identified that 16.1% of all hospitalizations were for ACSC. This is higher than what was found in previous studies, due to the inclusion of more conditions. Previous studies identified 4.4%20, 9.9%15 and 12.3%19 between 2012 and 2015. These studies used different lists, therefore the differences between results.
The development of a validated list for the Portuguese context is important, as the use of a common methodology can standardize results, enhancing comparability within the country. The use of the country-specific list can also better reflect the health system organization and population characteristics of Portugal, therefore with a higher specificity to the Portuguese context. It does however hinder international comparability. Having validated an ACSC list for the Portuguese setting it is expected that henceforward official authorities and academic research use this list when measuring ACSC hospitalizations.
Finally, it is important to address some limitations inherent not only to this paper, but also to the concept and operationalization of ACSC as an indicator of access and quality of care.
The use of experts’ opinion in the definition of ACSCs is subject to several biases, as well as the process of translating conditions into diagnosis codes. Furthermore, the use of administrative databases intended for financial purposes and subject to coding quality variations also recommends precaution in the interpretation of the results herein described. However, these variations should not be relevant enough to compromise the big picture, but should be taken into account when trying to zoom in to lower levels of aggregation. For example, while at the national and regional level this indicator is useful to identify Public Health priorities it might not be adequate at the local and individual level to evaluate specific providers’ access and quality once the aforementioned biases may cause an important lack of specificity. The regional and local use should also take into account several determinants of ACSC hospitalizations, such as socioeconomic status, hospital distance and rurality29 − 32, disease and multimorbidity prevalence24, in order to achieve a necessary risk adjustment of results.