Chapter 45 — The Iatrogenic Loop

When treatment changes memory, cognition, behaviour or emotion and the treatment-altered presentation is then interpreted as evidence of underlying illness, allowing intervention itself to become part of a self-reinforcing diagnostic loop.

When Treatment Becomes Part of the Evidence ↑

There is a point at which observation becomes intervention.

A clinician does not simply watch.

They prescribe.

Sedate.

Restrain.

Admit.

Discharge.

Change medication.

Change the environment.

Sometimes those interventions are necessary.

Sometimes they help enormously.

But every intervention also has the potential to change the person being observed.

Then another problem appears.

What happens when the change produced by treatment is interpreted as evidence of the illness which supposedly required the treatment?

Iatrogenic ↑

Medicine already has a word for harm caused by medical care.

Iatrogenic.

The word does not automatically mean negligence.

It does not automatically mean malpractice.

A properly prescribed treatment can still produce an adverse effect.

A necessary operation can produce a complication.

A useful medicine can create an unwanted reaction.

The existence of an iatrogenic effect tells us something about causation.

It does not by itself tell us whether somebody acted wrongly.

The Prescribing Cascade ↑

There is also an established medical concept called a prescribing cascade.

A medicine is prescribed.

The medicine produces an adverse effect.

The adverse effect is mistaken for a new medical problem.

Another medicine is prescribed to treat it.

The second medicine may then create another adverse effect.

And the chain can continue.

The concept is well recognised, particularly in older people taking multiple medications.

The important mistake occurs near the beginning.

The adverse effect is not recognised as an adverse effect.

It is reclassified as illness.

The Diagnostic Version of the Same Problem ↑

The loop I am interested in can occur even when no second medicine is added.

Treatment changes the person.

The changed person is assessed.

The change is attributed to an underlying disorder.

That interpretation is written into the record.

The record then justifies further treatment.

The treatment produces further change.

The changed person appears to confirm the record.

I call that an iatrogenic loop.

The Medication Is Also in the Room ↑

A person may look confused.

But confusion has causes.

A person may look sedated.

Sedation has causes.

Memory may fail.

Memory failure has causes.

Behaviour may become unusual.

Behavioural change has causes.

Before assigning those observations to personality, dementia, mental illness or character, there is a basic question:

What has changed?

Medication belongs in that question.

Before and After ↑

Chronology can be extraordinarily useful.

What was the person's baseline?

What treatment was introduced?

When was it introduced?

What changed afterwards?

Did the dose change?

Did the symptoms change with the dose?

What happened when the medicine was reduced, changed or stopped under appropriate medical supervision?

Did the person move back towards their previous baseline?

Those questions do not prove causation by themselves.

They create a much better investigation.

Dechallenge ↑

Pharmacology even has a useful word for part of this.

Dechallenge.

What happens after the suspected medicine is removed or reduced?

Does the adverse effect lessen?

That can add evidence to a causal assessment.

The corresponding idea of rechallenge — giving the drug again and seeing whether the effect returns — may provide information in some circumstances.

But deliberately reproducing a harmful reaction can be unsafe and unethical.

This is not an invitation for patients to stop or restart medicines on their own.

It is an argument for clinicians to treat medication timing as evidence.

Lorazepam ↑

This question became immediate for me again much later in life.

I was given lorazepam.

What followed, as I experienced it, included a striking deterioration in my ability to form and retain new memories.

I also experienced changes in behaviour which felt profoundly unlike my usual baseline.

The effect was frightening partly because memory was already a sensitive area in my life.

I knew what dissociative gaps felt like.

This felt different.

The Drug Information Matters ↑

Lorazepam is a benzodiazepine.

Its recognised adverse effects include sedation, amnesia, memory impairment, confusion and disorientation.

Paradoxical behavioural reactions can also occur.

Older adults can be particularly vulnerable to adverse effects from benzodiazepines.

That does not establish that every change I experienced came from the drug.

Other causes still have to be considered.

But it means that medication effect cannot responsibly be ignored as an explanation.

The Dangerous Reversal ↑

The dangerous reversal occurs when a recognised possible drug effect is observed and then attributed entirely to the patient.

The medicine can produce memory impairment.

The patient develops memory impairment.

The impairment is then used as evidence that the patient has a more severe cognitive disorder.

The drug disappears from the explanation.

The patient remains.

That is the reversal which interests me.

The Side Effect Becomes Evidence Against the Patient ↑

Once that happens, an adverse effect can acquire a second life.

The immediate effect may disappear.

The written interpretation may not.

“Confused.”

“Poor memory.”

“Behavioural disturbance.”

“Agitated.”

“Disinhibited.”

If those words enter the record without the treatment context, they can later appear to describe the person's ordinary state.

A temporary treatment effect can become a permanent biographical fact.

What Was the Baseline? ↑

This is why baseline matters so much.

How was the person functioning before the intervention?

Were they managing their own affairs?

Remembering appointments?

Writing?

Using computers?

Holding complex conversations?

Keeping records?

Recognising people?

Making decisions?

A snapshot taken after medication should not erase the person who existed before it.

Medication Is Not the Only Possible Cause ↑

There is another danger here.

Once I became alert to medication effects, I could make the opposite mistake.

I could blame every change on the medicine.

That would be another premature explanation.

Confusion can arise from infection.

Low oxygen.

Sleep deprivation.

Pain.

Dehydration.

Metabolic disturbance.

Neurological disease.

Delirium.

Psychological distress.

An unfamiliar environment.

Medication.

Withdrawal.

Or several of those together.

The lesson is not:

Always blame the drug.

It is:

Do not remove the drug from the investigation merely because a diagnosis is available.

Older People Are Particularly Exposed ↑

The problem becomes especially important in older people.

They are more likely to have several medical conditions.

More likely to take several medicines.

More likely to encounter multiple clinicians.

More likely to have changes in kidney or liver function which alter how some medicines are handled.

And more likely to have symptoms such as dizziness, falls, confusion or fatigue attributed to ageing itself.

The more complicated the person becomes medically, the easier it can be for the medication list to disappear into the background.

Polypharmacy Makes Causation Harder ↑

One medicine is comparatively easy to think about.

Five are harder.

Ten are harder again.

A new symptom may be caused by one drug.

An interaction.

A changed dose.

A changed organ function.

A newly added over-the-counter product.

Or the illness itself.

That complexity is not an excuse to stop asking.

It is a reason to ask more carefully.

The Informed-Consent Problem ↑

There is another layer.

Consent to treatment is meaningful only if the person has a reasonable understanding of what they are consenting to.

What is the medicine for?

What are the important risks?

What alternatives exist?

How long is it intended to be used?

What should happen if cognition or behaviour changes?

Who will review it?

A treatment which can itself alter memory, judgment or behaviour creates a particular responsibility to get those questions right before the effect occurs.

Consent Can Become Harder After the Drug ↑

There is an uncomfortable paradox.

A person may have the greatest capacity to question a sedating or cognition-altering medicine before it is administered.

Afterwards, the medicine itself may reduce their capacity to remember what was discussed, articulate an objection or reconstruct what happened.

That makes documentation before treatment especially important.

Who Notices the Change? ↑

The person themselves may notice.

A family member may notice.

A nurse may notice.

A psychologist who knows the person's baseline may notice.

Someone reading only the current chart may not.

Continuity matters because change can only be recognised against something.

Without a baseline, deterioration can look like identity.

When the System Has No Memory of the Baseline ↑

Institutions often have excellent records of intervention.

Medication charts.

Observation charts.

Incident reports.

Progress notes.

But the person's ordinary functioning may be much less visible.

The institution therefore possesses detailed records of the altered person and very little documentation of the person before alteration.

That imbalance can become consequential.

The Behavioural Description Can Outlive the Chemistry ↑

A drug is metabolised.

The concentration falls.

The adverse effect may diminish.

But the progress note does not metabolise.

It remains.

It may be copied.

Summarised.

Imported into another service.

Read years later.

The chemistry leaves the body.

The interpretation remains in the record.

A Prescribing Cascade Can Become a Narrative Cascade ↑

The established prescribing cascade adds medicine to medicine.

There can also be something like a narrative cascade.

An adverse effect produces behaviour.

The behaviour produces a description.

The description produces a diagnosis.

The diagnosis influences the next observer.

The next observer produces another description consistent with the first.

Soon the narrative has accumulated apparent support.

The original drug effect may have disappeared from view.

The Intervention Can Protect Its Own Explanation ↑

This is the loop at its most dangerous.

The treatment is justified by the diagnosis.

The treatment alters the patient.

The alteration is taken as confirmation of the diagnosis.

The confirmed diagnosis justifies more treatment.

Every turn appears to strengthen the starting assumption.

Unless somebody asks:

What if the treatment is helping to produce what we are seeing?

There Are Other Ways the Loop Can Begin ↑

Medication is only one way an intervention can alter the person and then alter the explanation of that person.

There are at least two other parts of my experience which belong here.

One concerns the movement from observation into hypothesis, proposed treatment and anticipated support.

The other concerns continuity.

What happens when care is repeatedly handed from one practitioner or team to another?

When a Working Hypothesis Begins to Behave Like a Fact ↑

There were times when I experienced another kind of loop.

An idea about what might be happening could begin as a working possibility.

Treatment could follow.

Expectations could follow.

Proposed future support could follow.

Yet very little might be explained to me about what was established, what was provisional and what was simply being considered.

A hypothesis can be useful.

Medicine could hardly function without them.

But a hypothesis should remain recognisable as a hypothesis.

Otherwise the patient can begin living inside an anticipated future that has never actually been established.

I experienced occasions on which support was spoken about as though it would be needed and available, only to discover later that what had been spoken of did not have the pathway I had understood it to have.

That can create its own kind of harm.

The expectation becomes real to the patient even when the proposed support never becomes real in the system.

I need to return to the records before describing those episodes in greater detail.

The Rotational Roster Effect ↑

The second problem took me longer to understand.

I came to think of it as the rotational roster effect.

That is my description, not a medical diagnosis.

A patient can remain in the same bed while the people responsible for their care repeatedly change.

The incoming practitioner inherits a record.

They may have only a limited period in which to assess the patient, understand what has already been tried and decide what they themselves should do.

Then another person arrives.

Continuity can become much harder than activity.

A Few Minutes in a Lift ↑

Some of my understanding of this came from an unexpectedly brief conversation.

I was in a lift with two sets of paramedics and other hospital staff.

There was banter between them.

The subject turned to short rosters and the way successive practitioners could approach a patient's care.

One of the people in the lift gave a description which immediately connected several things I had been wondering about.

As I remember it, the point was roughly this:

When somebody comes onto a short rotation there can be pressure to be seen to have done something.

Changing or adding a medication is highly visible.

The paperwork shows activity.

It can be easier to record than the slower work of spending time with a patient and deciding that the best action may be to leave a successful plan alone.

Another person in the lift responded that it was the best description of the phenomenon they had heard.

My recollection is that the others around us recognised what was being described.

The entire exchange probably occupied no more than a minute or two.

It has stayed with me ever since.

A Conversation Is Not a Study ↑

That lift ride did not prove that every medication change made during a rotation was unnecessary.

It did not establish an institutional policy.

It did not establish the motives of any particular doctor.

It gave me something else.

A hypothesis.

A possible explanation for a pattern I had already been trying to understand.

Afterwards I became more alert to the sequence.

Who had just taken over?

What changed?

Why did it change?

Was the previous plan failing?

Had the patient been asked?

Was the reason documented?

Did the next practitioner understand why the previous practitioner had made the change?

Transitions Are a Known Safety Problem ↑

The broader continuity problem is not merely my observation.

Medication safety guidance treats transitions of care as a recognised risk point.

When responsibility moves between clinicians, teams or settings, information can be lost.

Medication lists can diverge.

Reasons for previous decisions can disappear.

A change which made sense to one practitioner can arrive in front of the next practitioner without its history.

That is why medication reconciliation and communication at transitions matter.

Change Is Not the Same as Care ↑

There is a cultural problem hidden inside this.

Doing something is visible.

Not doing something can look passive.

But sometimes careful medicine consists of deciding not to interfere.

A medication may already be working.

A dose may already be appropriate.

The patient may already understand the routine.

The safest intervention may be continuity.

If a system rewards visible change more readily than thoughtful continuation, activity can begin to masquerade as quality.

Over-Treatment Can Harm Too ↑

Under-treatment can cause harm.

So can over-treatment.

More medication is not automatically better medicine.

More tests are not automatically better investigation.

More intervention is not automatically better care.

Every addition carries another possibility of adverse effects, interaction, confusion or loss of understanding about what is actually helping.

The appropriate amount of treatment is not the maximum amount.

It is the amount justified by the patient's condition, evidence, goals and informed agreement.

The Agreement Has to Survive the Roster ↑

There is another form of continuity which matters.

An agreement made between a patient and one clinician has little value if the next person does not know it exists.

The purpose of a medicine.

Its intended duration.

A target oxygen range.

A lifting restriction.

A plan to review rather than intervene.

A patient's refusal.

A patient's consent.

Those things have to survive handover.

Otherwise the patient may be forced to renegotiate the same boundary every time the roster changes.

For Me, Duration Mattered ↑

In my own experience, two examples became particularly important.

One was lorazepam.

Duration mattered.

Why was I receiving it?

For how long?

Who was reviewing the effect?

What was supposed to happen if my memory or behaviour changed?

The second was oxygen.

There, rate and target mattered.

Oxygen sounds so inherently beneficial that it can be difficult to think of it as another treatment requiring a dose.

But it is.

Oxygen Is Also a Treatment ↑

In some respiratory conditions, including COPD, oxygen has to be controlled and titrated rather than simply increased without limit.

Too little oxygen can be dangerous.

Too much can also be harmful in susceptible patients.

The treatment therefore has a target.

That makes communication about the target part of the treatment itself.

The Intervention in the Lift ↑

I remember another lift journey for a different reason.

An attendant stood up, came to my side and adjusted my oxygen.

I remember him telling me what he was doing.

The exact words and the precise setting belong on my CHECK list.

What I remember clearly is the sequence.

He intervened.

The oxygen was adjusted.

My breathing settled fairly quickly afterwards.

That does not prove by itself what had caused the difficulty.

But the response to the change was information.

It deserved to remain attached to the event.

Improvement Is Evidence Too ↑

This returns me to something I wrote earlier in this chapter.

There is something especially troubling when improvement after removal or adjustment of an intervention is treated as irrelevant.

If a person becomes markedly more capable after a suspected cause is removed, reduced or corrected, that change deserves attention.

It may not settle causation.

But it belongs in the causal investigation.

Improvement is not an inconvenience to the original explanation.

It is evidence.

The System Can Produce Movement Without Producing Continuity ↑

That may be the larger lesson from the roster problem.

A great deal can happen to a patient.

Orders can change.

Medicines can change.

Plans can change.

Staff can change.

Notes can accumulate.

Every page can demonstrate activity.

Yet the patient can experience the whole process as discontinuity.

The question is not only:

How much was done?

It is:

Did the people doing it understand the sequence, the previous decisions, the patient's baseline and the patient's agreement?

Resolving the System's Difficulty Is Not the Same as Resolving the Patient's Need ↑

There is another form of movement which can look like resolution from the institutional side while leaving the patient's original problem untouched.

A difficult need appears.

The service cannot readily meet it.

Medication may enter the picture.

Or discharge.

Or transfer.

Or another placement.

The immediate operational problem has changed.

The patient's need may not have changed at all.

I do not have direct evidence that these responses were deliberately used to avoid meeting my needs.

I do have repeated events which make the distinction worth preserving.

Medication does not provide psychological support which was never supplied.

Transfer does not answer a complaint which was never resolved.

Discharge does not make an environmental trigger cease to have occurred.

A new address does not automatically create continuity.

A system can resolve its difficulty without resolving the patient's need.

Medication and Movement Can Also Change the Evidence ↑

There is a second reason this matters in an iatrogenic loop.

Medication can change cognition, memory, affect or behaviour.

Movement can remove the person from the environment in which the original problem was occurring.

Either can make the next observation look different.

That difference may then be treated as though it explains the original event.

Sometimes it may.

Sometimes it may merely show that the conditions have changed.

The distinction matters because an intervention should not be allowed to erase the question which made the intervention necessary.

What Changed After I Left Hospital? ↑

There was another kind of before-and-after comparison becoming visible while I was writing this chapter.

I had moved from hospital to interim care and then to NewDirection Care.

Here I felt much more relaxed.

I felt better in myself mentally.

I was again managing much of my own day.

I remained mostly room-bound because of my physical limitations, but I did not experience that as a need to withdraw from people.

I could sit in the communal area for meals when I wanted to.

Most of the residents around me were living with substantial cognitive decline, so the opportunities for the kind of extended conversation I enjoy were limited.

That was not the same thing as feeling distressed by my surroundings.

The stress which had accompanied much of my earlier care had reduced.

My own observation was that my mental health had improved with it.

That does not prove which part of the change produced the improvement.

It does mean the improvement belongs in the record.

Improvement After a Change Is Still Information ↑

Medicine quite properly pays attention when a person's condition deteriorates.

It should also pay attention when the person improves.

What changed?

Was a medicine removed?

Was an environmental stressor removed?

Was there greater autonomy?

Was there less conflict?

Was sleep better?

Was the person managing more of their own routine?

Was a treatment no longer being imposed in the same way?

None of those questions should be answered merely because one answer would suit an existing diagnosis.

The improvement itself is part of the evidence.

The Handover Had Changed More Than My Address ↑

Looking back across the moves, I also realised that some treatment changes had travelled with me without my being clear about the decision which had created them.

Oxygen was one example.

Before that period I had generally used oxygen as required.

During the transition from hospital that became, as I understood it, continuous oxygen at two litres a minute.

At the time I was not functioning well enough to interrogate the change in the way I might normally have done.

The oxygen came with me.

The explanation did not travel nearly as clearly.

That is another form of discontinuity.

When a Temporary Decision Becomes the New Baseline ↑

A hospital decision may be entirely reasonable for the circumstances in which it is made.

The difficulty comes when the circumstances change but the intervention continues because it has become the inherited setting.

The next service sees:

oxygen at two litres.

It may not see:

why two litres,

when it began,

what the earlier baseline was,

whether it was intended to be temporary,

or what would trigger reassessment.

A setting can quietly become a fact.

I Began Asking the Question Again ↑

Once I became more settled, I began wondering whether continuous oxygen was still the treatment I required.

I trialled some periods of sleep without it and noticed that on some mornings I felt better and more capable for several hours afterwards.

That interested me.

It did not answer the medical question.

A few mornings of feeling better cannot tell me what my oxygen or carbon dioxide levels were while I slept.

Nor can they establish the flow I should use during the day or during exertion.

What they can do is justify asking for the treatment to be reviewed.

I still knew that exertion could create a very different respiratory demand.

So the question was not:

Can I prove that I no longer need oxygen?

It was:

What do I need now?

Review, Monitor and Coordinate ↑

That became the more sensible direction.

Review it.

Monitor it.

Coordinate it.

I wanted a plan which distinguished resting oxygen from exertional oxygen and nocturnal oxygen.

I wanted to know the target rather than merely inherit a flow rate.

I wanted the reason for the treatment to remain attached to the treatment.

And I wanted whatever was agreed to survive the next handover.

That is not resistance to treatment.

It is an attempt to make treatment intelligible.

The Same Question Applies to Medication ↑

The same reasoning brought me back to my medication list.

My instinct has long been that, where it can be achieved safely, less unnecessary medication is preferable to more.

I would put that more carefully now.

My aim is not zero medication at any cost.

It is no medication without a current purpose.

What is this for?

What benefit is it producing?

What risk does it carry?

Is the original indication still present?

Is the dose still appropriate?

Could it be reduced?

Could it be stopped safely?

Would stopping it create a greater risk?

Those are questions for a medication review, not decisions to make by simply removing tablets.

The Treatment Should Have to Keep Earning Its Place ↑

Illness changes.

People change.

Environments change.

Risks change.

Goals change.

A treatment which made sense six months ago may still be essential.

It may also no longer be necessary.

The fact that it appears on yesterday's medication chart does not answer that question.

Treatment should remain reviewable.

It should continue because the present evidence supports it, not merely because the previous roster handed it forward.

Not Every Iatrogenic Effect Comes From a Drug ↑

The word iatrogenic can draw attention towards medication.

But the broader problem is not confined to chemistry.

The treatment environment can contribute.

A handover can contribute.

A failure to communicate an agreement can contribute.

A failure to respond to an identified trigger can contribute.

An interpretation can contribute.

A sequence of individually understandable decisions can combine into something harmful for the patient.

That does not mean every difficult hospital experience is iatrogenic.

It means the possibility of harm created by care itself should remain inside the investigation.

That Question Is Not Anti-Medicine ↑

Questioning an adverse effect is not rejection of medicine.

It is medicine.

Pharmacovigilance exists because treatments can cause harm.

Medication review exists because prescriptions need reconsideration.

Deprescribing exists because a treatment which was once useful may become unnecessary or harmful.

Adverse-event reporting exists because unexpected outcomes matter.

Good medicine contains mechanisms for discovering when medicine itself is part of the problem.

Nor Is Every Cascade Wrong ↑

There is an important qualification.

Not every prescribing cascade is inappropriate.

Sometimes a medicine is essential.

Its adverse effect is known.

Another treatment safely manages that adverse effect.

The overall benefit still outweighs the risk.

The problem is not that medicine sometimes treats the consequences of medicine.

The problem is failing to recognise what is happening.

Recognition Changes the Ethical Position ↑

If the adverse effect is recognised, choices become possible.

Continue because the benefit justifies it.

Reduce the dose.

Change the timing.

Switch medicines.

Treat the adverse effect.

Stop the medicine safely.

Monitor.

The patient can participate in those choices.

Misattribution removes that possibility.

The person is treated for a disease they may not actually have instead of being told that treatment itself may be contributing.

The Importance of the Word “May” ↑

This chapter needs that word.

May.

A drug may have caused the change.

An illness may have caused it.

Both may have contributed.

The evidence may remain incomplete.

I do not improve medicine by replacing one certainty with another.

The purpose is to keep causation open long enough to investigate it.

A Simple Sequence ↑

If I had to reduce the whole chapter to one practical sequence, it would be this:

What was the baseline?

What changed?

What happened immediately before the change?

What medicines or interventions were introduced?

Are the observed effects known consequences of those interventions?

What other explanations fit?

What happened when the intervention changed?

Was the treatment context preserved in the record?

Those questions do not solve every case.

They make some preventable mistakes harder to make.

The Patient Should Not Have to Escape the Treatment to Be Believed ↑

There is something especially troubling when improvement after removal of an intervention is treated as irrelevant.

If a person becomes markedly more capable after a suspected cause is removed, that change deserves attention.

It may not settle causation.

But it is evidence.

The patient should not have to remain impaired for the original interpretation to remain credible.

What Medicine Can Learn From Its Own Effects ↑

An adverse effect is not merely an unfortunate interruption to treatment.

It is information.

It can tell us something about the medicine.

The patient.

The dose.

The interaction.

The diagnosis.

Sometimes it tells us that the original explanation needs to change.

That is not failure.

Failure begins when the information is made to disappear.

Then the Record Takes Over ↑

The immediate medical event may last hours.

Days.

Weeks.

The record can last decades.

A treatment-altered state can be compressed into a few words.

Those words can travel farther than the explanation of how they arose.

The next clinician may never meet the untreated person.

They meet the record first.

And that brings me to the next problem.

What happens when the record becomes more powerful than the person?