GAD-7 is one of the most commonly used methods for identifying probable cases of generalised anxiety disorder. It’s gained prominence thanks to its ease of use, with patients able to complete the test on their own. However, you might be wondering how reliable GAD-7 is at correctly diagnosing this condition.

In this article, we’ll put GAD-7 to the test by checking its sensitivity and specificity. We’ll also explain a little bit more about why these are such useful metrics for assessing reliability, providing examples along the way.

What is GAD-7?

GAD-7 is a self-report questionnaire for anxiety. More specifically, it screens patients generalised anxiety disorder, and also indicates the severity of the condition. It was first developed in 2006 by a team led by Robert Spitzer. In their article A Brief Measure for Assessing Generalized Anxiety Disorder, they noted that there had previously been “no brief clinical measure for assessing GAD”.

As the name suggests, GAD-7 consists of seven questions. Patients are asked to think about the following aspects of their mental health over the preceding two weeks:

• Nervousness or anxiousness
• Inability to stop worrying
• Worrying too much
• Having trouble relaxing
• Restlessness
• Irritability
• Feeling afraid

Patients then score themselves based on how often they have experienced each of the problems listed above. If they haven’t had to deal with a certain problem at all, they give themselves a score of 0. A problem that came up on several days garners a score of 1; for more than half the days, the score is 2; and for nearly every day, the score is 3.

After answering all of the questions, the patient’s score is added up. If it is 10 or higher, this indicates a probable case of GAD. Similarly, the higher the patient’s score, the more severe their anxiety can be said to be, with scores of 5–9 representing mild anxiety, 10–14 representing moderate anxiety, and 15 or above representing severe anxiety.

GAD-7 is not intended to be used as the sole means of diagnosing generalised anxiety disorder. Instead, further evaluation is recommended for patients with a score of 10 or higher, along with potential changes to treatment.

GAD-7 example

To show how GAD-7 works, we've put together an example below.

	Not at all	Several days	More than half the days	Nearly every day
Nervousness or anxiousness	0	1	2	3
Inability to stop worrying	0	1	2	3
Worrying too much	0	1	2	3
Having trouble relaxing	0	1	2	3
Restlessness	0	1	2	3
Irritability	0	1	2	3
Feeling afraid	0	1	2	3
Column totals	0	2	2	3

In this table, the total scores obtained in each of the columns are 0+2+2+3, giving a grand total of 7. This means that this example patient could be said to have mild anxiety, but does not meet the cut-off point for a diagnosis of GAD.

GAD-2

GAD-2 is a sub-scale of the main GAD-7 questionnaire. Instead of including all seven questions, the patient is only asked the first two: how often they have felt nervous, anxious or on edge, and how often they have been unable to stop or control their worry. The cut-off score here is 3, at which point further evaluation is recommended.

Measuring reliability: Sensitivity and specificity

In an ideal world, we would have 100% accurate tests for every health condition. Unfortunately, this isn’t possible in reality. There will usually be a number of cases that a test isn’t able to detect—or, conversely, the test may incorrectly diagnose some healthy people as having a certain condition.

To help us to determine whether a test is accurate enough to correctly diagnose most people, we calculate two values: sensitivity and specificity.

Sensitivity refers to the likelihood that someone who does have a certain condition will test positive for it. A very sensitive test will pick up almost all cases, which makes them useful for ruling out that condition. That’s because it will rarely diagnose a sick person as being healthy.

Meanwhile, specificity refers to the likelihood that a healthy person will test negative, meaning that they don’t have the condition being tested for. A very specific test will correctly classify these people as being healthy. As such, it can help you to rule in a certain condition, because it will rarely diagnose a healthy person as being sick. We can’t expect tests to be perfectly specific and sensitive. However, according to Power, Fell and Wright’s Principles for high-quality, high-value testing, a test’s sensitivity and specificity added together should be at least 1.5. In other words, if you measured sensitivity and specificity in percentages, then you should get at least 150% when adding these two together.

How to calculate sensitivity and specificity

To explain how these concepts work, imagine a hypothetical disease—we’ll call it Exampleitis. We’ll also imagine that there’s a test to check whether you have Exampleitis, which we’ll call EX-7. To help us illustrate things, we’ve created the image below.

On the left hand side of this image, we can see a group of 10 people with Exampleitis. Meanwhile, on the right, there’s a group of 10 people who don’t have the disease.

The story doesn’t end there, though. All 20 of the people in the image took the EX-7 test to determine whether they had Exampleitis. Anyone who got a positive result is shown in green, while anyone who got a negative result is shown in pink.

If EX-7 were completely accurate, then the 10 people on the left would all be in green, and the 10 people on the right would all be in pink. However, no test is completely perfect, and, as you can see, there are some errors.

Sensitivity

On the left hand side of our image, we can see that 8 of the people who had Exampleitis got positive results. These are known as true positives. 2 of them, however, got negative results even though they also had the disease. These are known as false negatives.

This side of the image is where we work out the sensitivity of our EX-7 test. We do this by dividing the number of true positives by the total number of people who have the disease—including both the true positives and the false negatives. By performing this calculation, we can work out that the test managed to diagnose 8 out of the 10 people with Exampleitis. This means that its sensitivity is 80%.

Specificity

The right hand side of our image shows those without the disease. Here, 9 people correctly received negative results—true negatives. Meanwhile, 1 person was incorrectly identified as having Exampleitis. This was a false positive.

Using this side of the image, we can work out the specificity score. To calculate this, we need to divide the number of true negatives by the total number of people without the disease. Here, we can see that there were 9 true negatives and 1 false positive. As such, the specificity score of EX-7 is 90%.

Our example here is, of course, very simplified. After all, in the real world, the sensitivity and specificity of a test wouldn’t rely on so few results. However, it enables us to easily see how these concepts work, and to understand how they might apply to GAD-7.

How reliable is GAD-7?

The NHS describes GAD as potentially “difficult to diagnose”, partly because it can result in symptoms similar to those of other mental health conditions such as depression. It makes sense, therefore, to ask exactly how reliable any test for GAD can be.

To determine the reliability of GAD-7, we can take a look at the data from Spitzer et al. In their initial study, they approached 2,982 subjects, of whom 2,739 completed the form. Based on the results obtained, the research team tried a range of cut-off scores to determine the best possible balance between sensitivity and specificity.

Their conclusion was that a score of 10 was a good cut-off, as at this point, both sensitivity and specificity were above 80%, making GAD-7 a useful test for generalised anxiety disorder. More precisely, 89% of patients with GAD had scores of 10 or above, while 82% of those without GAD had scores of less than 10.

As such, the sensitivity of GAD-7 is 89%, while its specificity is 82%. This gives us a combined score of 171% or 1.71, satisfying the criteria for a useful test.

Spitzer et al also tried a higher cut-off score of 15 for GAD-7. Here, they found that while specificity was maximised, the test’s sensitivity was only 48%. In other words, while virtually all positive test results were likely to be correct, setting the cut-off so high would likely miss a large number of cases.

Further research

Since the initial 2006 study, GAD-7 has come into widespread use in care settings. Other researchers have conducted further studies to determine the usefulness of the questionnaire both for diagnosing GAD and for other anxiety-related conditions.

In a literature review by Rutter and Brown, a variety of different studies have since shown that there is “strong psychometric support for the GAD-7 in a variety of contexts”. A study by Delgadillo et al, for example, found that a GAD-7 score of 9 and above “had a sensitivity of 80% and specificity of 86% for any anxiety disorder”.

However, some studies have suggested potential limitations to the use of GAD-7. A study by Kertz, Bigda-Peyton and Björgvinsson found that, while GAD-7 “appears to be a good measure of generalized anxiety symptoms”, it is less effective at screening for GAD due to poor specificity. These findings were backed up by a later study by Beard and Björgvinsson, who commented that there was a “high false positive rate”.

Using GAD-7 for other conditions

Though GAD-7 was created to test for generalised anxiety disorder, it has come to prove moderately useful at testing for certain related conditions. According to research by Nerys Williams, these include:

• Panic disorder — sensitivity 74%, specificity 81%
• Social anxiety disorder — sensitivity 72%, specificity 80%
• Post-traumatic stress disorder — sensitivity 66%, specificity 81%

In addition to the above figures, the National HIV Curriculum also found that GAD-7 acts as a reasonably good screening tool for any anxiety disorder, with a sensitivity rate of 68% and specificity of 88%.

Diagnosing GAD

GAD-7 can prove efficient at screening for GAD and determining the severity of an individual’s anxiety. A formal diagnosis, however, requires the input of a medical professional, who will assess the patient against a set of criteria.The two most commonly used criteria are DSM-5, which was produced by the American Psychiatric Association, and ICD-11, from the World Health Organisation. They share many similarities; a brief summary of both can be found below.

Medical professionals may check for a range of different symptoms, including headaches, gastrointestinal symptoms and insomnia. They may carry out a physical examination in which they check for signs such as increased heart rate, trembling or a shortness of breath. The patient is also likely to be asked about their medical history. For more information about how GAD is diagnosed, read the NICE guide.

Reliability of GAD-7: Conclusions

When gauged against usual benchmarks, GAD-7 stands out as a test with very good levels of specificity and sensitivity. This enables it to serve as a useful screening tool for generalised anxiety disorder, as well as helping to ascertain how severe a patient’s anxiety is.

Nonetheless, further studies remind us that all tests have their limitations, and that while a score of 10 or above on GAD-7 can prove a good indicator of an anxiety disorder, this does not represent a definitive diagnosis.

If you have taken the questionnaire yourself and think you may be above the cut-off point for GAD, consult a medical professional for a formal diagnosis. They’ll help you to access the treatment you need to live a fuller, happier life.