(1) General guidelines of the Portuguese screening programme
In 1998, the Portuguese General Health Department (DGS) has established the first guidelines for DR population-based screening programmes. Non-mydriatic Chamber Fundus Photography (colour retinography) was the recommended screening method, due to its high sensitivity and specificity (92% and 90% respectively), and because this method can be performed by trained paramedical personnel and later sent for ophthalmologist analyses. Annual screenings were recommended for diabetics after puberty .
Regional Health Administrations (ARS) have the responsibility of operationalize population-based screening programmes. In Portugal there are five ARS (ARS North, Central, Lisbon and Tagus Valley, Alentejo and Algarve). So, since 2001, ARS began the implementation of screening strategies under DGS guidelines [35-39].
None the less, the guidelines were vague in what concerns to major operationalization aspects as what services and health staff should be involved and which are their responsibilities, where the screening test should take place, who identifies and convokes the diabetic populations, etc. Therefore, the strategies adopted by each ARS are significantly different .
In 2018, DGS issued new and more detailed, guidelines for the organization of regional screening programmes , proposing a flow chart for the screening process (Figure 2).
(2) Regional DR screening protocols
At the North Regional Health Administration (ARSN), the DR Screening Programme began in 2009, and has been gradually implemented in the following years. In 2009, ARSN, developed exhaustive proceedings, documentation, and protocols, which have been subsequently expanded and adjusted .
In this region, retinographies are performed in Primary Health Centers. However, there are no fixed retinographers in health facilities. The equipment remains in mobile units, moving from health center, to health center, according to prior established schedule . Primary Health Centers are responsible for identifying and convening the diabetic population and retinographies are performed by orthoptics technicians. However, there are no orthoptics technicians dedicated solely to the screening programme. Those professionals are provided by local hospitals, and usually accumulate the functions inherent to the screening programme, with the functions they perform regularly in the hospitals. After the retinographies are performed they are analysed and graded by ophthalmologists . ARSN is conducting a research aiming the introduction of automatic image reading software in DR screening programme, however, this technology is still experimental . After the grading, positive cases are referred to the hospital for treatment. The ARSN uses a specific software to support the screening programme (SIIMAScreenings) .
At the Portuguese Central Region Health Administration (ARSC), the DR Screening Programme is running since 2001 . As in the North Region, the screening method and the target population follow the 1998 DGS guidelines . Until 2011, the screening protocol was similar to the one implemented at ARSN. However, in that year, was introduced the use of an automatic image reading software (RetmarkerSR) in conjunction with the traditional human analysis and grading. This software allows the detection of RD lesions such as DME and small haemorrhages in retinal photographs, through a method based on image processing algorithms . Two of the selected papers focus on the performance of this particular software revealing a sensitivity of 99.76% and a specificity of 99.49% [43, 44].
Another particularity of ARSC Screening is that there is no software application to support the screening programme. The data are requested by the ARSC to each of the Primary Health Center Clusters (ACES) and compiled into Excel sheets .
Lisbon and Tagus Valley Regional Health Administration (ARSLVT) and the Association for the Protection of Diabetics of Portugal (APDP) signed a cooperation protocol in 2009, for DR screening . It was the beginning of Diabetic Retinopathy Screening Service for Lisbon and Tagus Valley (RETINODIAB), commissioned and driven by APDP and supported by ARSLVT. The RETINODIAB follows the 1998 DGS norms in terms of screening test and target population .
In 2016, the ARSLVT implemented their own pilot screening in four ACES. Accordingly, with this established protocol, the retinographies are performed by orthoptists, in the ACES, and automatically analysed and graded by a software - “Retmarker". When classified by the software as "necessary human reading", they are sent for ophthalmologists’ analysis. The results of these readings are made available to the family doctor by means of a computerized screening platform. Positive and inconclusive cases are referenced to hospital ophthalmology services .
Nowadays, ARSLVT, extended this new screening programme, and APDP, RETINODIAB, is still a complementary response, continuing to cover seven of the fifteen ACES .
In ARSLVT, the screening programme is computer-supported by SIIMAScreenings in 4 ACES and by the APDP system in 7.
An internal recruitment process for orthoptists for Primary Health Care has begun in 2017 .
At Alentejo Regional Health Administration (ARS Alentejo), there is no standardized screening strategy. In fact, there are three different screenings. The DR screening managed by ARS Alentejo, which began in 2011 and follows 1998 DGS guidelines in terms of method and target population, is implemented in one ACES. The retinographies are performed by orthoptic technicians provided by hospitals and uses SIIMAScreenings as screening computer-system .
In a second ACES, family doctors refer patients with diabetes to perform the retinography in the hospital, so the data related to this ACES are not introduced in the screening platform. And, in a third area the screening is carried out in partnership with APDP .
In March 2013, the Algarve Regional Health Administration (ARS Algarve) began the implementation of a population-based screening for all diabetics in the region .
The screening test is performed by the two Hospitals in Algarve, in the ophthalmology departments. The articulation between ARS Algarve and the hospitals is performed through protocols and annual contracting. Screening monitoring is computer-supported .
During the year 2014, hospitals were reticent about the renewal of the screening protocol due to the reduce installed capacity. So, ARS Algarve proposed to limit the screening, in this period, to the "new cases" diagnosed during 2013 and 2014 what was accomplished by the end of the year (36). In 2015 and 2016, the screening was resumed in a normal way. However, in 2017 and 2018, the screening did not take place. In that year’s activities report, ARS Algarve claims that, although the normal procedures for the renewal of the programme were carried out, there was any hospital response and that, despite having taken countless efforts to develop a screening programme less dependent on hospital capacity (similar to those existing in the North, Center and part of Lisbon and Vale do Tejo), this was not possible due to numerous procedural constraints .
The analysis of the technical documentation of the five ARS, showed that there are considerable differences between the implemented screening programmes (Table 2) [35-39]:
- The screening location varies: in ARSN and ARSC there are portable retinographers which, in turn, are allocated to the Primary Health Centers of the region [39, 46]; at ARSLVT there are fixed retinographers in Primary Health Care units , and in ARS Algarve all screening phases are performed by hospital ophthalmology services.
- If, in some ARS, retinographies are performed by hospital orthoptic technicians, which accumulate the functions in the hospital with the DR screening, other (ARSLVT) are hiring optometrists for primary health care units. Although this solution seems simple and effective on eliminating the dependence of available hospital technicians, it is not easy to implement, mostly due to the lack of consensus on the competence of optometrists to perform retinographies. In fact, there are substantial differences in the training of the two types of professionals: orthoptic technicians are qualified to detect vision abnormalities and ocular motility disorders. Therefore, the orthoptic technician is active in diagnosis, therapy and rehabilitation; on the other hand, optometrists are the professionals that, through examination of the eye, diagnoses refractive errors and prescribes appropriate lenses and/or exercises, without the need for drug or surgical treatments.
- In the ARSC and in part of the ARLVT region, artificial intelligence software is implemented for automatic retinographies grading. Several studies state its acceptable sensitivity and specificity levels and its effectiveness to reduce ophthalmology services burden.
The new DGS directives substantiate an important attempt to guarantee quality, equity of access and standardization of screening at national level . However, the analysis of the latest activity reports of the ARS (2018), clearly shows that, so far, the new guidelines have not produced effects at the regional level. Thus, while some ARS established procedures perfectly framed with the guidelines now issued, there are others, in which the so-called population-based screening programmes fall far short of the requirements that the denomination, and the current national guidelines, require.
(3) Main indicators and screening results
The analysis of the official reports of the Portuguese institutions directly involved in the implementation of the DR screenings allowed to determine a set of common indicators, used to monitor the process and the results of the screening programmes.
However, the number of available indicators is very small, reflecting only the concern with the coverage of the screening [35-39]. No indicators inherent to the quality of the process  were found in any of the five ARS. In rare cases, references to the evolution of the number of positive DR cases were found, which, however, were discarded due to important inconsistencies in the concept of "positive case" itself. Still, it was found that most ARS collect and report the following indicators:
As previously mentioned, generally, the indicators are calculated by the ARS, although the data are obtained directly through an operating system dedicated to screening, or indirectly, through requests to the primary health units, or associations involved (APDP, hospitals) Of course, when the second case occurs, less reliability of the data is expected, since it is common for different entities to follow different criteria for extracting and pre-processing the information.
But, in addition to this issue, there are other inconsistencies in the calculation of the indicators [35-39]:
1- First, as we have seen, there are several ARSs (part of ARS LVT, ARS Alentejo and ARS Algarve) where screening is still conducted, in whole or in part, by other institutions, leaving the question of whether it is truly a population-based screening. Normally, the ACES where this happens are counted as being covered by a screening programme, but, at the risk of, in some cases, providing only an opportunistic screening to registered diabetics.
This inconsistence will affect the “Geographic Coverage” indicator.
2- The variable "number of identified diabetics" is also likely to introduce some bias in the analysis of the results. In reality, not all identified diabetics are convolved into screening. According to the DGS 'own guidelines, family doctors should remove from the list the subjects who are unable to remain seated, those who underwent a retinography less than a year ago and those who are blind. Thus, it is important to distinguish whether the ARS account for the initial number of identified diabetics, or that obtained after the purging of the initial listings.
The “Population coverage” and “Screened population” indicators could be affected by these decisions.
3- The variable “Number of invitations” is also not easy to measure. In fact, so far, none of the ARS has managed to strictly comply with the 12-month interval between screenings. Therefore, at the time of the change of civil year, there are several locations with the annual screening still in progress. Thus, these questions arise: is it effective only to consider invitation letters in places where the screening has already been completed? All invitation letters sent should be considered, even if, in some cases diabetics have not yet had the opportunity to adhere to the screening, simply because, the screening was scheduled for a date later than the present moment? The assumptions in each case are not clear and may condition the comparison of adherence rates between ARS.
The “Population coverage” and the “Adherence rate” are affected by this bias.
Despite the constraints mentioned previously, the following tables 3, 4, and 5 show the available indicators, in each of the five ARS. Due to the scarcity of information in some of the ARS, it was decided to present the results only for 2015 and 2017 (years in which more comprehensive information was obtained) (39). The variable "Number of retinographs performed" was the only one that allowed an evolutionary analysis, which is presented in table 5 [35-39].
Despite several setbacks in all regions, the number of screenings has been increasing since 2009. In 2015, a total of 113,443 retinographies were taken, 19% more than in the same period of 2014 (Table 5). However, access to diabetic retinopathy screening is still remarkably variable in Portugal and needs urgent attention. Population coverage, in 2017 varies from 0% in ARS Algarve to 100% in ARSLVT (Table 4) [35-39].