Niahealth is a proactive health service that specialises in the early identification of health risks and recommending strategies to mitigate them. We primarily focus on lifestyle modification strategies but our clinicians do sometimes recommend medications such as statins if applicable. 

Our interpretations of test results and our recommendations have a real impact on our user’s lives which is a responsibility that we take seriously. We are committed to ensuring that any test or intervention that we recommend is based on the best available evidence. In this section we cover Niahealth’s approach to evidence.

‍

Two key questions 

Two questions drive our assessment of any test or intervention: (1) Is it safe?; (2) Is it effective?

Any medical intervention be it a laboratory test, a medication, or surgical procedure must have a clear safety and efficacy profile before it can be recommended. They are the fundamental questions for safe clinical practice.

Exactly how you answer these questions depends on the case. For a blood test the direct safety concerns may appear to be minimal and related only to the blood draw. It is far more important in this case to determine whether that test is going to be effective. Will it provide accurate and reliable information that can be used to make sound decisions? If the test is likely to be inaccurate then there is a risk of providing false or misleading information with a potentially significant impact on an individual’s health. (A false negative on a cancer screening test would be a good example). Safety and efficacy are therefore intimately linked in medical testing.

Safety concerns are much more obvious when recommending interventions. All interventions have potentially harmful side-effects. This is most obvious for medications and it’s why the first trials that a medication goes through focus purely on establishing a safe dose. Even behavioural interventions like diet and exercise are not without risk. For example, high-intensity interval training regimes should not be recommended for someone with a known risk of cardiac arrhythmia. We need to know when we can safely recommend an intervention and when it should be avoided. We also need to know what potential complications need to be monitored. 

‍

Assessing the quality of evidence 

In order to determine the safety and efficacy of a test or intervention effectively we need to choose good quality evidence. In medical school we are taught a hierarchy or “pyramid” of evidence to help with this choice. The study types at the base of the pyramid are the least “generalisable” or reliable and the studies at the top are where you can theoretically have the most confidence in the results and apply them to individuals. 

The least reliable form of evidence is “expert opinion”. Medicine often involves dealing in grey areas where research will not tell you exactly what to do. In these cases a panel of experts may be consulted to give their best guess based on their experience - the idea being that this is better than individual clinicians trying things out on their own.

Next come the “observational” studies which, as the name suggests, involve collecting data on people to try to answer a question without doing any particular intervention. Case reports are observational studies on a single, or perhaps a handful, of individuals. These might throw up interesting results but they simply don’t contain enough data to be able to draw reliable conclusions for everyone. They can be improved on by collecting data from larger populations and separating them into groups to compare based on what you are interested in. 

Observational studies with comparison groups are where we find a lot of the research into longevity for reasons detailed below.

‍

The top half of the pyramid starts with trials. These are intervention based studies and are how new treatments are assessed for their efficacy. A study population is split into two groups, one of which will receive the treatment, and the other will receive something to compare to, like a placebo or the current best treatment for a condition. In essence these are experiments to determine whether something actually works in a particular group of people. The best trials will include randomisation of participants to reduce bias and control as many factors as possible. These are the closest we can come to proving an effect in humans. 

The best source of evidence in this hierarchy are systematic reviews. These are large and comprehensive analyses which combine data from multiple other studies. There are set protocols for conducting a systematic review which include requirements to explain why studies were included or admitted and report on the overall quality of the evidence. The best systematic reviews will interrogate the data in the studies they include to identify any potential sources of bias or inaccuracy before providing a conclusion.

‍

The challenge 

The pyramid makes it seem simple. If you need to answer a question you should look for systematic reviews and randomised controlled trials for your answers. Unfortunately, as with much of what you learn in medical school, the truth is not so simple. 

When we research longevity or preventative medicine we are looking to answer the question “what should we do to stay as healthy as we can for as long as we can?”. This is not the question that the majority of biomedical research looks to answer. The vast majority of studies are focussed on studying particular diseases to find new treatments. There is nothing wrong with this (although many will say there is) it just isn’t the kind of evidence we are looking for. 

Longevity research has another issue which makes robust randomised control trials difficult. We are interested in what will happen to health over decades rather than years. Generating RCT evidence over that timescale takes prohibitive amounts of time and money (although newer measures of aging such as biological age clocks may improve this). So in the absence of RCTs we need to step a little further down the pyramid and use observational studies. There are many more of these and some of them are of very high quality. The National Health and Nutrition Examination Survey (NHANES) for example is a huge study based in the US that provides detailed health and behavioral data on thousands of participants. This has been interrogated in hundreds of studies to try to find which behaviors and interventions change the risk of death or other outcomes. Using high quality studies like these also allows different risk profiles of particular groups to be teased out in the hope that interventions can be targeted.

We can assess the quality of these observation based studies by looking at their sampling, their statistical methods, and the size of the effects they have found. Fundamentally though these will only ever give us correlations of various certainty - not causation.

‍

From the population to the individual

Niahealth is not a research institution. It is a clinic that helps real people which generates a well recognised tension in medical science. Our evidence is all at the population level and presents assessments of risk but none of it can tell you exactly what will happen to the person in front of you. This is where judgement is key.

The key to effectively translating population level research to the individual is to understand the evidence that you have and communicate it honestly. Much of the research we can call on is more applicable to certain individuals than others. Study populations are historically biased towards white males living in western nations. Different ethnicities, geographies, genetic backgrounds, and crucially females have all been systematically neglected in medical research which significantly affects our ability to provide clear recommendations. 

These baked in sources of bias are why the most important part of our approach to evidence is to point out where it is missing. This allows us to have open discussions with individuals about how they should look after their health based on what we currently know. 

‍

Our approach to choosing tests

In plain language, there are two key considerations about clinical tests related to the safety and efficacy discussed above. 

1. The test needs to reliably tell you something about the risk of a disease. 
2. You need to be able to do something about it 

Answering point one is usually a simple case of due diligence in the literature. All clinically validated tests must report their sensitivity and specificity which determine the risk of false negatives and false positives respectively. The important nuance to bear in mind though is that these metrics depend on a particular situation or “pre-test probability”.

The pre-test probability is an estimate of the likelihood of a positive test before the result is known. In traditional clinical practice this would be determined by criteria like the patient’s age and sex as well any findings from assessing the patient. Pre-test probability is a challenge when conducting screening tests, as Niahealth does, and it leads us to be more stringent about the tests we will include. We avoid using tests that are only indicated for diagnosing specific conditions after a full assessment by a clinician since they would not be appropriate in our care delivery model. 

Point two is more about practicing medicine well than about the research. In general, there is no point doing a medical test unless you will receive information that could make you healthier. Put differently, the test needs to give an answer that leads to a specific recommendation. There are thousands of possible tests out there but not all are appropriate to every situation. We have exercised our clinical judgement to determine a list of tests that provide actionable information for proactive medicine. 

‍

Final perspectives 

Niahealth exists in an evidence light space within medicine which presents significant challenges to evidence based practice. We can’t rely on randomised controlled trials and systematic reviews to tell us what to do which means we have to take a very close look at the evidence that does exist. In this effort there are a few key principles that we can follow:

1. Focus on safety and efficacy and pause if there is insufficient evidence to clearly answer either question. 
2. Where applicable, stay in line with regulated medical bodies (e.g. colleges). This is particularly relevant when recommending pharmacotherapy. 
3. Appraise evidence carefully and identify sources of inaccuracy and bias.
4. Be honest with individuals about where the evidence is lacking so that they can make appropriate informed decisions about their health. 

With those principles in mind, the goal of the research we do at Niahealth is to provide our clinicians with the information they need to follow those principles with confidence and practice safely. Ultimately the practice of medicine always comes down to a clinician’s judgement on individual cases based on experience and evidence - Niahealth is no exception to that.  

In this we are inline with this summary of “evidence based medicine” from the British Journal of Medicine: “Evidence-based practice is the “integration of best research evidence with clinical expertise and patient values.” It means that when health professionals make a treatment decision with their patient, they base it on their clinical expertise, the preferences of the patient, and the best available evidence. We apply the same principles when we choose a new car, a restaurant at which to dine, or film to watch at the cinema. To find out whether they are good or not, we read restaurant reviews, watch specialist car programmes, ask friends, and refer to past experience.”The line “best available evidence” brings me to the final important point. Evidence in medicine is always evolving as new studies are reported. Truly evidence based practice means retaining enough openness to change what you do when new compelling evidence emerges.

‍

Sources

1. Golden, Sherita & Bass, Eric. (2013). Validity of Meta-analysis in Diabetes: Meta-analysis Is an Indispensable Tool in Evidence Synthesis. Diabetes care. 36. 3368-73. 10.2337/dc13-1196. 
2. https://www.england.nhs.uk/tis/wp-content/uploads/sites/17/2014/09/tis-guide-finding-the-evidence-07nov.pdf