Background

The Brazilian Unified Health System (SUS) was created in 1990 to implement the 1988 Brazilian Constitution health component, which established universal health access, including all aspects, as a fundamental right. ¹ Although part of a unified system, health information systems (HIS) are historically part of specific verticalized control programs like malaria, tuberculosis, and hepatitis, which developed as independent initiatives in the past. Therefore, health information used to be fragmented into distinct databases. ^{Reference Coelho Neto and Chioro2}

An important landmark happened in 1975, when the Ministry of Health (MoH) promoted the first meeting on HISs to discuss the implementation of the main HIS. ^{Reference Jorge, Laurenti and Gotlieb3}

Thereafter, Brazilian HIS have substantially increased in quality, mainly in the last few years with the migration to web-based services enabling fast and secure data transfer from municipalities to the MoH’s servers and providing access to summary reports. These HIS allowed data collection in a patient-level granularity, allowing researchers, health professionals, and policymakers to conduct ecological and individual analyses. ^{Reference Jorge, Laurenti and Gotlieb3}

Among those HIS, we emphasize the Information System of Diseases Notification (SINAN), which stores data from all mandatory notification diseases, the Information System of Live Births (SINASC), responsible for collecting and storing data from all live births, and the Information System of Mortality (SIM), where death registers are collected and stored. Others HIS were developed for specific needs, such as SIVEP malaria (used to collect and store data of malaria infection in the Amazon region) and SIVEP gripe, used to collect and store data of severe acute respiratory syndrome in the whole country.

Despite the considerable advances in the last years, those HIS still need to overcome some critical issues, like the absence of a primary key to enable data from the same person to be linked quickly from the different databases. More essential, difficulties on broad Internet access poses an additional barrier to the timely availability of data so that epidemic detection can be performed in real time. Additionally, MoH’s open data policies are still unclear, sometimes only a byproduct of specific research projects. For instance, despite the efforts of the MoH to make COVID-19 data available during the pandemic, delays due to both the data entering and data availability are still observed. A chain of events in the notification process, which are also sources of delay, has posed an additional obstacle to monitor the pandemic in real time. ^{Reference Bastos, Economou and Gomes4}

Responses to epidemics require the ability to detect signs and symptoms quickly, even if the pathogen is unknown. The existing HIS are not thought to allow prompt epidemic detection because of their nature and limitations. ^{Reference Bastos, Economou and Gomes4}

In recent years, medical records are increasingly moving toward electronic platforms. These clinical registers are a rich source of patient information and epidemiological information, allowing for timely data, in-depth investigation of patient condition, follow-up, and sample availability for further analysis. ^{Reference Ludwick and Doucette5}

Here, we propose a complementary approach by using EMRs data collected from health units in real time to generate a user-friendly interface with visual analyses to enable insights from the local health surveillance system personnel. Details on the whole process can be viewed in Figure 1.

Figure 1.

Flowchart of surveillance process based on electronic medical records.

Case

As a case description, here we present a dashboard and respective findings from São Caetano do Sul, São Paulo. ¹⁰

Data from a hospital and a medium complexity health unity from SCS were assessed. In 2022, from January to September, almost 120,000 medical appointments were performed in the two health facilities, with a daily average of almost 4,000. Out of the total registers, 11,400 had mention of “fever” in the patient complaint register. Most of the population were women with ages ranging from 20 to 49 years.

In the whole period, the most frequent terms mentioned when fever was also detected were pain in the throat, cough, headache, and coryza, with little variation within the last 6 months. The fever rates ranged from 1,050 to 1,025 per 1,000 inhabitants, and the highest were in Fundação and Centro neighborhoods, in the northern region of the city (Figure 2A). The fever rates, higher than the neighborhood’s population, can be explained by both the quality of the SCS health system and the proximity to São Paulo, the Capital.

Figure 2.

(A) Fever incidence rate per neighborhood (per 1.000 hab.), in São Caetano do Sul. (B) Time series of case counts relating “fever” as a major symptom and positive COVID-19 cases confirmed by RT-PCR in The Corona São Caetano Platform (CSCP).

Figure 2B shows the time series of fever relating cases, which reveals a bimodal pattern, probably describing the Omicron BA.1 outbreak (weeks 1–6) and the Influenza A outbreak (weeks 20–30) through January–February 2022. The main diagnostic hypotheses in the medical records of the first epidemiological weeks of 2022 pointed out the ICD J111 (Influenza – flu – with other respiratory manifestations, caused by unknown viruses), followed by B342 (infection by unspecified coronavirus). These findings raised flags to the overlapping outbreaks of both COVID-19 Omicron BA.1 subvariant and Influenza A virus, which were confirmed by our team by analyzing data from private laboratories and CSCP. ^{Reference Leal, Mendes-Correa and Buss6,11}

Final remarks

Synergy between private sector health units and governments must be stimulated to the wide use and implementation of such systems of EMRs. Therefore, companies developing those systems could fill the gap of harmonizing and structuring databases for EMRs. To do so, the government should provide guidelines on those database structures. Timely information generated from EMRs will be a very important tool to the decision-making process as well as research in epidemiology. Quality and security on the data produced is of paramount importance to allow the use by health surveillance systems.