risk – AQuAS Blog

Improving the quality of healthcare in intensive care units. The PADRIS programme in the Tarragona Datathon 2018 (part two)

14 Feb

Today we continue the interview with Maria Bodí (@mariabodi23), doctor at the Intensive Medicine Service, and Josep Gómez (@JosepGomezAlvarez), doctor in Biotechnology at the Hospital Universitari de Tarragona Joan XXII, experts in clinical management and aspects of quality and safety in healthcare.

In your opinion, what are the conclusions of the so-called Real World Data studies which were carried out using real data in terms of benefits and risks related to patient safety?

Benefits? Everything. Information derived from a real healthcare environment is necessary to take decisions. Randomised clinical trials which have defined the effectiveness and safety of therapeutic interventions have served till now as a Gold Standard of the best scientific evidence. But they are very costly, and what is more, they are aimed at very selective groups of patients. Studies and analyses derived from the real world, known as Real World Data, mean it is possible to know the effectiveness and safety of interventions in groups of patients which are usually excluded from trials (pregnant women, older people, patients with many comorbidities who are the majority, etc…).

There is a series of limitations and obstacles that prevent Real World Data from substituting clinical trials. On the one hand, legal and ethical aspects and those concerning the guarantee of the quality of data. On the other, it is not possible to accept the bias that there is when not randomising Real World Data. Before taking decisions, we need to ensure that there are no confusing factors.

Real World Data complements the information obtained from clinical trials in routine clinical practice.

Some weeks ago we received the visit of Lucian Leape in Barcelona, author of the extremely famous book “To Err is Human”. We were fortunate to be able to listen to him in a talk at AQuAS – were you able to listen to him?

No, unfortunately we were not able to attend. It is a pity because we were told that it was a good review of what we have learnt in the past two decades in the field of clinical safety and that some good recommendations were made for the future.

In your opinion, what contributions did the To Err is Human report, published 20 years ago, make?

It revolutionised the field of safety. It was a paradigm shift which is still in place today.

The To Err is Human report denounced the thousands of deaths in the United States resulting from adverse events which could have been avoided! People who were dying in hospitals for reasons unrelated to the disease for which they had been admitted. The most important thing is that those events, those deaths, could have been avoided. Better training, better work organisation, knowing and analysing risks and teamwork, among other factors, have been shown to be key elements that contribute to reducing the amount of events and their severity.

What impact has the PADRIS programme had on your day to day?

In 2016, MIMIC-II was published, a huge set of de-identified data of ICU patients of the Harvard Medical School. It was created and has been maintained by researchers at the Massachusetts Institute of Technology. They published this dataset with the aim of democratising research. The idea is that after attending a training course on the treatment of data for research, you are certified as a practising researcher and you sign a document of usage: ultimately, to enable a researcher to have access to a large data base to carry out research. In addition, it encourages researchers to share the code (data processing methodology) that they have applied to data to obtain the results they publish. Altogether, it makes studies more transparent and repeatable which in turn increases the excellence of scientific production.

As a result of our experience in extracting data from the clinical information system to develop our management tool of the unit, we took on the challenge of generating our own data base to carry out research. Once created, we contacted the PADRIS programme so they could advise us regarding data anonymisation protocols and methodologies to get access to the programme in order to carry out research. They showed great interest in our project at all times and they helped us see it to fruition and thus the role of the PADRIS programme was decisive in making the Datathon Tarragona 2018 possible. We are in fact still in contact with them to define future strategies about how to make this data available for research projects without infringing any data protection law and how to broaden the database with data from other ICUs in the Catalan territory.

If you could make a recommendation to other researchers who wish to do research, what would it be?

We recommend they collaborate with experts in other fields, especially those related to data and statistical technology. We are now reaching a level of sophistication and volume of data that obliges us to work in multidisciplinary teams in order to make the most of data and to understand it the best we can. The datathons are a great example of this; the role of the clinician is decisive in defining aims and in validating the results that appear when cutting edge algorithms are applied by data scientists. At the same time, the role of data scientists is also decisive when suggesting and applying complex methodologies which are far removed from traditional statistics applied within a clinical context.

What professional challenge would you like to succeed with in 2019?

The ultimate professional challenge for 2019 is the same as each year: improving the care given to patients which are admitted to an ICU. To achieve this, we have some very specific challenges in our unit. On the one hand, to continue developing our tool to exploit data which enables us to analyse processes and the results obtained in our day to day and in this way become aware of where we need to place our attention to make improvements. On the other hand, taking advantage of the secondary use of data to carry out research and to generate algorithms for automatic learning which are able to help a doctor take the most accurate and appropriate decisions based on the profile of each disease.

(You can read the first part of this interview here)

Improving the quality of healthcare in Intensive Care Units. The PADRIS programme in the Tarragona Datathon 2018 (part one)

7 Feb

Last November, the AQuAS played an important role by means of the PADRIS programme in the Critical Care Data analysis Summit and Tarragona Datathon 2018. Talking about it has been in the pipeline since then.

The PADRIS programme contributes to the improvement of people’s health by making anonymised health information available for re-use by researchers in Catalan research centres, in compliance with the current legal framework and established principles.

Is this seen in practice based on a professional’s experience?

Today we interview Maria Bodi (@mariabodi23), doctor of the Intensive Care Medicine service in the Hospital Universitari de Tarragona Joan XXII, expert in clinical management and aspects of quality and safety in healthcare. Like many health professionals, she combines healthcare practice with research.

What is your day to day like?

As head of the Intensive Care Medicine Service at the hospital, in my day to day my basic task focuses on managing the service and organising the care for critical patients by coordinating the work of the professionals involved. More than 150 people work in the service including medical professionals (specialists in intensive medicine, resident intern doctors), nurses, nurse aids, ancillary staff, physiotherapists and secretaries. In addition, it is a service which participates and collaborates in the teaching of medical, nursing and physiotherapy degrees or certifications.

I try to facilitate the participation of professionals in the strategy of the service. This requires articulating and coordinating all efforts made with a clear objective in mind, which is providing quality care to our patients. To a greater or lesser degree, it is necessary to encourage and coordinate the participation of professionals in care-giving, management, teaching and research. This will guarantee the commitment of a worker in the service’s strategy and its organisation.

If we focus on medical professionals, each member of the team is in charge of a specific area of our speciality and we therefore provide ongoing training for the entire team, assess results and commit ourselves to carrying out actions which derive from the analysis of our results.

What do you think about the format that was used for the Datathon?

The Datathon was the result of a series of developments over recent years within the field of secondary use of data in patients’ clinical records for management and for research at a top level. The experience was very good. Pure science, with doctors, technicians and technology all at the service of real-life data analysis in order to find the best scientific evidence.

In the last three years, our group has delved deeper into the study of data and also into the assessment of the quality and safety of data for secondary use. Our progress has allowed us to collaborate with other leading teams such as the team of Dr Leo Celi of the Massachusetts Institute of Technology with whom we organised this event.

In your opinion, how can the quality of care be improved in intensive care units?

We have to aim for excellence. We need to tackle all the dimensions regarding quality of care. We need to improve effectiveness, safety and efficiency. But when talking about good and efficient results, we are not referring to the number of actions done at a particular cost. We are talking about bringing value to the patient, to the work team, the organisation, the health system and society. How is this achieved?

Our team has worked on developing a methodology which enables us to have automatic indicators of quality. This has been possible because all of a patient’s bedside devices (mechanical ventilation, monitoring, dialysis machines, etc.… ) are connected to the system of clinical information and this information, from the hospital’s clinical records, from laboratories and imaging tests, is integrated. This information is incorporated by professionals in an ordered manner.

By means of these indicators designed with innovative technology, it is possible to assess the processes of care and their results. If the professional participates in the design of a process of care, in its planning, and knows how things are being done and what results there are, they get involved and commit themselves to the aims of the service and the organisation.

Likewise, one needs to move from a reactive to predictive medicine, preventive and personalised. We have data with which to start working along these lines. Nevertheless, the scale and complexity of this data makes it difficult for methods of artificial intelligence to easily translate it into relevant clinical models. The application of cutting-edge predictive methods and data manipulation require collaborative skills between professionals who are medical and technology experts, as well as new models in the treatment and analysis of data.

We have read that it is possible to assess risk in an intensive care unit in real time. It seems difficult to imagine this. What can you tell us about it?

It can indeed. We work on the assumption that we have access to the data stored from all patients that have been admitted to an ICU, or in more than one ICU. If the combination of a group of variables (demographic, clinical, laboratory results) leads to a complication or an adverse event, a trained computer model using this data can predict the risk of the same complication or event occurring if it detects the combination of this group of variables. This is the basis of predictive medicine.

Understood, but in practical terms, what is purpose of assessing the possibility of a risk?

It can range anywhere from analysing a risk or predicting the appearance of a complication in the course of a disease, a problem of safety, an adverse event, the need for or an increase in dosage of a specific drug, to a specific therapy. It can predict the probability of an improvement or worsening and even the risk of death for a patient as a result of their disease.

Medicine already has calculators to assess a risk of death due to a disease based on data bases created by professionals using manual registers. But these days, with the automation of data registration- an example of this being the Clinical Information Systems in our ICUs – today’s methodology based on big data and artificial intelligence allows us to have much more detail when assessing risks.

When there is talk of moving towards a health model based on value, what does this mean exactly?

It is about organising work around the specific conditions of a patient which optimises their cure. It is a question of paying organisations and hospitals for the value they bring. Their results and their costs are the key components of the value which a health system and its professionals offer patients. But if we are talking about results, we are not only interested in whether a patient survives the ICU or not. Results are measured in terms of quality, their capacity of going back to their normal lives, their jobs, etc…

Therefore, to push efforts for improvement, we need to base ourselves on multidisciplinary work and a health model based on value; this means changing an organisation’s business model and investing in systems of measurement, analysis of clinical results and costs.

What importance do you attribute to the fact that data is obtained automatically and not manually?

It means data can be analysed using one source only, errors are minimised and a professional’s time is not required for the task of inputting the data.

How is the data included automatically?

In the ICU, clinical information systems now enable all information to be integrated. Apart from the data which professionals input in an ordered manner during the process of care of a critical patient, laboratory results, imaging tests, clinical records and the data from all a patient’s bedside devices are integrated too (mechanical ventilation, monitoring, dialysis machines, etc…).

What does “secondary use of data” mean?

Primary use is that which is used on a day to day basis, at the patient’s bedside, to take decisions relating to diagnosis, treatment and the planning of the process of care. For example, a decision is made whether to increase the dosage of a drug based on laboratory results.

We talk about secondary use when we refer to using data for management or for research. Obviously, the end aim is still the improvement of care for a patient.

(To be continued …)

Health and poverty are hereditary: can we remedy this?

17 May

In recent years, child poverty has increased in Catalonia as a result of the economic crisis. According to the 2016 figures from the Idescat, the latest figures available, and from 2009, children are the group most at risk of poverty, more than the adult population and also more than the 65-year-old or older population group.

“Child health and poverty. What can we learn from the data?” was the title of the conference held within the framework of the Celebration of the 2018 World Health Day.

Data from the latest report related to children and the effects of the crisis on the health of the population were highlighted at the conference, published by the Observatory of the Health System of Catalonia: children with a lower socio-economic level present up to 5 times more morbidity, consume more pharmaceutical drugs (three times more psychotropic drugs) than the remainder of the child population, visit mental health centres more frequently (5.9% of girls and 11.4% of boys as opposed to 1.3% and 2.2% in girls and boys with a higher socio-economic level) and are admitted more to hospitals (45 girls and 58 boys for every 1000 as opposed to 13 and 26, respectively) especially for psychiatric reasons.

A child’s health largely depends on the economy of their parents and those that belong to families with a lower socio-economic level have more health problems, a fact that can have disastrous consequences in other areas such as education and social life and which condition their future. This fact is exacerbated in the case of children with special needs or chronic diseases where their health suffers even more from the effects of poverty because in some cases their care requires specific products which families cannot afford.

This is one of the problems that we are facing right now. There is growing scientific evidence, both in biology and in social sciences, of the importance of the early years in life (including exposure in the womb) in the development of the capacities that stimulate personal well-being throughout the life cycle. Childhood is also a structural transmitter of inequalities, both from a health and socio-economic point of view. If nothing is done, boys and girls who belong to families with few resources run the risk of growing up into adults with worse health and a lower educational and socio-economic level than others.

What can we do? We can of course strengthen the social welfare state, with structural and institutional reforms which are more than ever necessary. Educational policy is fundamental, especially by reinforcing primary education, guaranteeing equal opportunities and putting the spotlight on those children in a disadvantaged situation. Once they are adult, active labour policies are also needed. And from health policies, despite their eminently palliative nature, primary and community care is particularly important as is guaranteeing care to all children.

Post written by Anna García-Altés (@annagaal).

Stratification and morbidity database (2n part)

31 Mar

(This post is the second part of this post)

A key element for completing the stratification of population in risk groups is information system. It’s necessary to have a database that integrates information collected from different health records and therefore, in order to meet this demand, the population morbidity database was created.

The point is that every day there is more and more emphasis on the need to provide patients with a comprehensive and integrated health and social care, but the analysis and evaluation of this care can not be carried out correctly with fragmented information systems, on the contrary: it must be done starting from the integration of the data these contain.

Population morbidity database structure

The population morbidity database is based on a system of related tables that pivot around the users table, which includes the main data of the insured patient (demographics or health status, to name two examples).

Currently, there are three more tables: the diagnostic, the contact with health services and the pharmacy, but this type of structure relatively easily allows incorporating both information from new records (outpatient clinics, dialysis, respiratory therapy, etc.) and new tables with other relevant information, such as results of clinical findings:

Figure 2: Structure and content of the population morbidity database. The clinical determinations table in gray is not yet implemented.

Taula d'assegurats

The population morbidity database integrates information from the following records:

Registro Central de Asegurados – (RCA) (Central Registry of Insured Patients) managed by public relations management of CatSalut. This register basically provides all the information of residence, socio-demographics and health status of the insured patient.
Registros del conjunto mínimo básico de datos – (CMBD) (Records of basic minimum data set) managed by CatSalut Division of demand and activity analysis. These records feed both into the diagnoses table and in the contacts tables. There are different registers to collect information from the healthcare lines:
Hospitalization (CMBD-HA): information provided by general acute care hospitals (hospital admissions, outpatient surgery, home hospitalization, day hospital) from 2005 to 2014
Socio – sanitary (CMBD-SS): information of the care provided by the health centres of internment (long and medium stay and UFISS) and outpatient care equipment (PADES) from 2005-2014.
Psychiatric hospitalization (CMBD-SMH): information of the care provided by psychiatric hospitals from 2005-2014.
Outpatient Mental Health (CMBD-SMP): information of the care provided by outpatient mental health centres for the period 2005-2014.
Primary Care (CMBD-AP) information on the care provided by primary care teams from 2010 to 2014.
Emergency (CMBD-UR): information of emergency care (hospital and CUAP) from 2013 to 2014.
Pharmacy activity Log (RAF) managed by CatSalut’ management of pharmacy and medicine unit. This record provides all the information about outpatient pharmacy dispensing for the period 2011-2014.
Record health services turnover (RF) managed by the Division of care services provision. This record provides information on any activity financed by CatSalut, but that does not rely on a specific record: hospital outpatient clinics, dialysis, home oxygen therapy, rehabilitation or non-emergency medical transport. This record provides information mainly on the contacts table for the period 2011-2014.

The possible uses of the population morbidity database are multiple: the population stratification, the specific analysis of certain health problems (broken femur, IC, COPD,…), the development of population indicators of efficiency in the use of resources, etc.

We can conclude that, for the volume of data that integrates, this database has the characteristics of a structured “big data”, with a considerable capacity for growth and adaptation to new requirements and data sources and offers enormous possibilities for analysis.

Post written by Emili Vela, head of Modules for Tracking Quality Indicators (MSIQ). Health Care Area. CatSalut.