The four things a strong recommendation always has
You do not need clinical training to check for these. Their absence is not proof that anything is wrong, but it is the fastest available signal, and asking about a gap costs you nothing.
1. A baseline you could measure again
Somewhere it should say what your child can do now, in terms specific enough that the same measurement could be repeated in three months by someone else. "Uses 15–20 single words spontaneously across a session" is a baseline. "Has significant expressive language delay" is a description.
Without a baseline, nobody can demonstrate later that anything changed — including in good faith. It is the most common omission and the most consequential.
2. Goals written as observable behaviour
A goal should describe something you could watch happening. "Will request a preferred item using a two-word phrase, with an unfamiliar adult, in four out of five opportunities" is checkable. "Will improve functional communication skills" is not — it can be declared met or unmet at anyone's discretion, indefinitely.
Look particularly for with whom and where. A goal that does not specify those can be satisfied entirely inside the therapy room, which is where skills go to look finished without being useful.
3. A stated intensity with a reason attached
The number of hours should have a justification next to it — this frequency, because of this evidence or this constraint. Not merely a number in a box.
Intensity is the figure most commonly inflated in this sector, and it is inflated in a completely ordinary way: not by fraud, but because more is safer for the clinician, more is reassuring for the parent, and more is better for the provider. Three pressures, all pointing the same direction, and nothing structural pushing back.
4. A review date and an exit criterion
When will this be looked at again, and what would have to be true for it to stop? A recommendation without an ending is a subscription, whatever else it says.
The exit criterion is the item most often missing entirely, and it is worth asking for in writing. The request itself tends to change the conversation.
What weak recommendations do instead
They describe the child rather than the plan
Pages of history and observation, then two lines of recommendation. It reads as thorough. It is not — you are paying for what happens next, and if that is the shortest section of the document, something is out of proportion.
They give a number with no unit of progress
"Twelve hours per week" with nothing saying what those hours are meant to achieve or when they would reduce. Ask directly: at what point would this go to ten? If nobody can answer, the number is not really a clinical judgement.
They recommend only what the author sells
A recommendation from a provider that concludes you need precisely the services that provider offers is not necessarily wrong. It is simply unverified. A good report will mention at least one thing the author does not provide — even if only to say it has been considered and ruled out.
They use volume as reassurance
Watch for intensity offered in response to worry rather than to evidence. If you express anxiety and the number goes up in the same conversation, that number is answering the wrong question.
Five questions to send back
Reply to any recommendation with these. They are ordinary, professional questions and a good provider will welcome them.
- What is the baseline for each goal, and how was it measured?
- At what point would you reduce this intensity, and what would trigger that?
- What would you expect to see in eight weeks if this is working?
- What would you expect to see if it is not, and what would you change?
- Is anyone's pay affected by how many hours this recommends?
That last one feels rude. Ask it anyway, of everyone, including us. A provider with nothing to hide answers it in one sentence — and the ones who bristle have told you something useful.
How to read the answers
You are not assessing clinical accuracy, which you cannot do. You are assessing whether the answers are specific and checkable. Specific answers can be wrong, and that is fine — a wrong specific answer gets corrected at the next review. A vague answer cannot be wrong, which means it also cannot be right, and nothing will ever be corrected.
Two evasions worth naming. "Every child is different" is true and is not an answer; the correct response is "understood — what would you expect for this child?" And "we'll know more as we go" is fine as a caveat and not fine as a plan; the follow-up is "what would you need to see, and by when, to know more?"
When the recommendation is probably fine
Most are. Plenty of clinicians in the UAE write careful, well-evidenced recommendations and would be glad to be asked these questions. The point of this guide is not suspicion — it is that you should not have to rely on luck.
If a document has baselines, observable goals, a justified intensity and a written ending, you are in good hands. Keep it, and hold the next review against it.