Why Reassurance Makes Health Anxiety Worse

A clean test result works. That is the part nobody says out loud, and it is why none of this yields to argument. The relief after a doctor says there is nothing seriously wrong is real, immediate and complete.

Then it drains.

In 1997 a team working in a British general hospital did the obvious thing that almost nobody had done, which was to keep measuring after the reassurance instead of stopping there. Sixty patients referred for gastroscopy were assessed before the procedure, immediately after a consultant told them there was nothing seriously wrong, and again at twenty four hours, one week, one month and one year. Worry and illness belief fell sharply for everybody at the moment of being told. In the group who had scored high on health anxiety beforehand, both had come back within twenty four hours, and a year later they sat where they started. Patients with moderate health anxiety kept what the reassurance gave them.

Same consultant, same words, same clean camera.

Sixty people is a small study and it never needed to be a large one, because the finding is not about how much reassurance helps. It is about how long.

Before any of the rest of it: people with health anxiety also get ill

This has to come early and without hedging, because everything after it can be misread as the claim that the symptoms are invented, and that reading is both false and dangerous.

A Swedish national cohort published in JAMA Psychiatry in 2023 followed 4,129 people diagnosed with hypochondriasis against 41,290 matched comparisons. Mortality in the hypochondriasis group was higher, not lower. All-cause hazard ratio 1.69. Death from natural causes 1.60. Mean age at death seventy against seventy five. The hazard ratio for suicide was 4.14, which should end any suggestion that this is a trivial or comic complaint.

Fear of illness has never protected anyone from illness, and it does not substitute for examination in either direction. In the largest treatment trial in this area, nine participants died during follow-up, all of pre-existing disease, six of them in the group that got standard care.

The word in the search box

Hypochondriac is what people type, and it is worth putting down before going further. It is an old word, it stopped being a clinical label a long time ago, and its main surviving use is as an insult between people who are not ill. Nothing it implies is accurate. This is not performance, it is not a bid for attention, it is not weakness, and it is not the same thing as worrying too much in general.

What it gestures at has a plainer description. Health anxiety is a pattern in two entirely ordinary processes, what attention lands on inside the body and what the mind then concludes, plus whatever gets done afterwards to settle the question. Peter Tyrer and colleagues screened 28,991 patients attending cardiology, respiratory, neurology, endocrine and gastrointestinal clinics in English hospitals and found significant health anxiety in 19.8% of them, rising to 24.7% in neurology. That is not an eccentricity. That is one person in five in the waiting room.

It is also, and this matters more than the prevalence figure, among the most treatable anxiety patterns clinicians see.

Reassurance has a half-life, and the half-life shortens

Here is the structure. It is worth reading slowly, because it inverts the obvious.

The reinforcer is the relief, not the information. When a question about the body gets closed, by a search or a check or an answer from somebody trusted, anxiety drops, and the drop is what the nervous system records. What it learns from the drop is that this was a category of question that requires closure. Which raises the resting alarm. Which produces more sensation, because a body under surveillance is a noisy body. Which generates the next question, sooner.

So the interval contracts. Early on, a clear answer buys weeks. Later it buys an afternoon. Eventually the reassurance has to be renewed inside the same conversation, which is the point at which people notice something is wrong and conclude they are getting worse. They are not getting worse. The loop is doing what reinforced loops do.

What is being chased underneath all of it is not information. It is certainty, and that difference is the whole reason the loop has no natural stopping point: examinations return probabilities, and no examination returns zero. The entry on intolerance of uncertainty is the long version of that sentence. Tolerance is the variable worth working on because it is the one that moves. The unanswerable question is not going to become answerable.

Both directions are the same loop

The version of this that gets missed entirely is the person who does not go.

DSM-5 divides illness anxiety disorder into a care-seeking type and a care-avoidant type, and in practice the division is much less tidy than that. In the sample discussed in a 2024 review of the disorder, 25% were care-seeking, 14% were care-avoidant, and 61% moved between the two.

Avoidance buys the identical relief on a slower schedule. Not booking the appointment closes the question by refusing to open it, the anxiety drops, and the loop is reinforced in exactly the same way. From outside it reads as indifference, sometimes as an admirable lack of fuss, which is why it goes unchallenged for years while the other version gets a reputation for being difficult. The cancelled appointment and the fourth opinion in a month are one fear running two tactics, and most people who have one have had spells of the other.

The practical consequence is that a person can appear to have solved this by force of will, having simply switched poles. Nothing about the loop has changed. Only the direction of the thing that closes the question.

What a symptom search is actually ranked by

Search results are not ordered by prevalence. They are ordered by what gets clicked, linked and written about at length, and that tracks how alarming a condition is rather than how many people have it.

Ryen White and Eric Horvitz named the consequence cyberchondria in 2009 and documented it in query logs: sessions that begin with a common complaint and escalate to a rare and serious disease. Headache to brain tumour is the canonical path. Their analysis of web content found the frightening explanation linked to the common symptom far more tightly than any base rate would justify.

Which means every symptom search performs a small inversion. It takes a distribution where the dull explanation holds almost all of the probability and hands back a list with the rare one on the first screen. Nobody is being lied to. The ranking is simply answering a different question from the one being asked.

Eurostat put the share of EU adults who looked online for health information at 60% in 2025, so this is not fringe behaviour. It is the default first move, and it is the only step in the loop that is free, instant and available at four in the morning. That combination is what makes it the hardest link to break.

Why asking the people closest to you costs the most

This is where the real damage happens, and it is the part described least honestly.

Brynjar Halldorsson and Paul Salkovskis interviewed people with obsessive compulsive disorder and with health anxiety about what reassurance seeking is actually for. One finding stands out. The health anxious participants did not describe seeking support at all. They described seeking resolution. Support means being accompanied while something stays uncertain. Resolution means extracting a verdict. From the doorway the two look the same, and they do opposite things to the loop, because a verdict is a reinforcer and company is not.

The faster and more confident the verdict, the worse the arithmetic. A partner who says of course it is nothing, I promise, closes the question hard. A hard close produces a large drop. A large drop is a large reinforcement, and a large reinforcement means a shorter interval. Loving someone well and answering them quickly are, in this narrow domain, at odds.

Then the second-order problem, which is the one that breaks relationships. Nobody can supply verdicts at that volume indefinitely. The answers get flatter, then slower, then withheld, and eventually there is a row about it. What the withholding produces, read from inside the loop, is not relief from the question. It is the precise conclusion the person was most afraid of. They have stopped telling me the truth. Something is being kept from me. The loop manufactures its own worst evidence out of fatigue it created itself.

The doctor half, done properly

Most advice at this point amounts to check less, which is close to useless as an instruction because it names a quantity and no mechanism. There is a better version and it is structural rather than motivational.

One doctor. The same one. With an agreed rhythm.

A relationship with a single clinician who holds the history does two things that nine appointments with nine strangers cannot. It builds a baseline, so a genuine change can be recognised as a change instead of being argued from scratch every time. And it makes a schedule possible, which is the real prize: an agreement that this gets looked at at an interval set in advance, in a calm room, rather than decided in the moment a sensation arrives. The checking still happens. It stops being a response to the body.

A systematic review of twenty two studies of continuity of care found significantly lower mortality with higher continuity in eighteen of them, with generalists and with specialists. That result was not produced in a health anxious population and it should not be stretched into one. It does establish something plain. Seeing the same doctor is not a soft preference.

What the evidence actually supports

Not a technique, and not from here. The relevant fact is narrower than advice and better than encouragement.

A systematic review of fourteen randomised trials found a large effect for cognitive behavioural therapy adapted to health anxiety, compared against waiting lists, treatment as usual, medication and other psychological therapies. Effect size d = 1.01 at the end of treatment, holding at six and twelve months. The CHAMP trial randomised 444 health anxious patients recruited from hospital medical clinics to between five and ten sessions or to standard care, and the advantage was still measurable eight years on. Therapy delivered inside general hospitals, to people who had turned up for a cardiology appointment rather than a psychological one.

Very little in this field has a record that clean. The reason to state it is not motivational. It is that health anxiety gets treated as a character trait to be endured more often than almost anything else in the category, and the evidence says it is a pattern that responds.

The health anxiety test reads the three parts separately, the noticing, the interpreting and what gets done about it, because they come apart and each points somewhere different. Fifteen items, one month, and a comparison against what other people report. It is not a diagnosis and it says nothing whatsoever about whether anybody is ill.

The first sign that the loop is loosening is not confidence about the body. It is a sensation that gets noticed, read as nothing in particular, and then forgotten by the evening.

Sources

• Lucock, M. P., Morley, S., White, C., Peake, M. D. (1997). Responses of consecutive patients to reassurance after gastroscopy: results of self administered questionnaire survey. BMJ, 315(7108), 572-575.
• Tyrer, P., Cooper, S., Crawford, M., et al. (2011). Prevalence of health anxiety problems in medical clinics. Journal of Psychosomatic Research, 71(6), 392-394.
• Mataix-Cols, D., Isomura, K., Sidorchuk, A., et al. (2023). All-cause and cause-specific mortality among individuals with hypochondriasis. JAMA Psychiatry, 81(3), 284-291.
• Kikas, K., Werner-Seidler, A., Upton, E., Newby, J. (2024). Illness anxiety disorder: a review of the current research and future directions. Current Psychiatry Reports, 26(7), 331-339.
• Halldorsson, B., Salkovskis, P. M. (2017). Why do people with OCD and health anxiety seek reassurance excessively? An investigation of differences and similarities in function. Cognitive Therapy and Research, 41(4), 619-631.
• White, R. W., Horvitz, E. (2009). Cyberchondria: studies of the escalation of medical concerns in web search. ACM Transactions on Information Systems, 27(4).
• Cooper, K., Gregory, J. D., Walker, I., Lambe, S., Salkovskis, P. M. (2017). Cognitive behaviour therapy for health anxiety: a systematic review and meta-analysis. Behavioural and Cognitive Psychotherapy, 45(2), 110-123.
• Tyrer, P., Cooper, S., Salkovskis, P., et al. (2014). Clinical and cost-effectiveness of cognitive behaviour therapy for health anxiety in medical patients: a multicentre randomised controlled trial. The Lancet, 383(9913), 219-225.
• Tyrer, P., Wang, D., Crawford, M., et al. (2021). Sustained benefit of cognitive behaviour therapy for health anxiety in medical patients over 8 years: a randomised controlled trial. Psychological Medicine, 51(10), 1714-1722.
• Pereira Gray, D. J., Sidaway-Lee, K., White, E., Thorne, A., Evans, P. H. (2018). Continuity of care with doctors, a matter of life and death? A systematic review of continuity of care and mortality. BMJ Open, 8(6).
• Eurostat (2025). Towards Digital Decade targets for Europe: internet use for seeking health-related information.