OET Reading practice test

1A patient arrives on a respiratory ward with a pulse oximeter clipped to her finger. While the nurse is still listening to her breathing, the screen has already produced a number. It is easy to see why that number travels so well. Someone can enter it on a chart, compare it with yesterday’s entry or repeat it over the telephone without the listener having seen the patient. The difficulty begins when a number that is easy to pass on is assumed to be equally easy to interpret. The patient’s effort in finishing a sentence does not fit as neatly into a box.

2A pulse oximeter estimates how much of the haemoglobin in arterial blood is carrying oxygen. It shines light through tissue and uses the changing signal with each pulse to make the estimate. Even a steady display is the end of a calculation. A poor signal, perhaps from a cold or moving finger, is one reason to doubt a figure. A more awkward possibility is that a device can settle on a plausible figure and still tend to read high under particular conditions. Warming a hand or repeating a measurement can address some problems. Agreement between two readings from the same instrument cannot establish that either agrees with the blood.

3In 2020, Michael Sjoding and colleagues used hospital records to ask a question that a table of average readings could easily conceal. Among people whose fingertip reading appeared to lie in a reassuring range, how often did a closely timed arterial blood sample tell a different story? They found such discrepancies more frequently in patients recorded as Black than in patients recorded as White. These were comparisons with a reference measurement for the same patient. Looking instead at whether one group had lower fingertip readings overall would have answered a different question, since the groups might also have differed in how ill they were.

4The records had a limitation that mattered for the next investigation. A racial category in a hospital record gives no direct measurement of the skin through which a device is taking its reading. People who share a category can have markedly different skin tones. Nor did the earlier comparison isolate the performance of any particular model under a common set of conditions. Its value was in making a pattern visible in ordinary care. To discover what happened as skin tone varied, researchers needed to measure it. Simply adding more records containing the same categories would have made the original comparison larger while leaving that gap open.

5The English EXAKT study took that step. Its investigators tested five fingertip models that had been used in home monitoring, but made their comparisons in intensive care, where arterial samples were available. A separate instrument measured skin tone. Across the five models, the estimated difference in displayed saturation between typical darker and lighter skin tones ranged from 0.6 to 1.5 percentage points at the same arterial saturation, with the higher reading for darker skin. That qualification about the arterial value matters. The comparison concerns what the instruments displayed when the underlying oxygenation was alike. It does not suggest that one group of patients had less need for oxygen.

6A difference of a single point can sound too small to deserve a meeting. Imagine, however, a service that reviews every reading on one side of a decision threshold. Two otherwise identical readings falling just either side of it will take different routes. A small shift can therefore matter greatly near that boundary while making little difference to a decision farther away. There is another temptation here: to subtract a fixed amount whenever a patient has a darker complexion. A group average cannot say whether a particular device is reading high in a particular patient, still less by precisely how much.

7There is a further wrinkle in the apparently straightforward label ‘home monitor’. In EXAKT the label describes where these models had been deployed, not the setting in which their performance was measured. The intensive care unit offered a reference sample that would rarely be available in a living room. That is a strength for a study of measurement. It also marks a limit. The researchers could compare device and blood readings closely, but they could not use those comparisons alone to tell us how often someone at home would call for advice, or what the person answering the call would hear.

8What, then, should a hospital ask before buying its next monitor? A figure summarising agreement for everyone is a start. More revealing questions concern who was studied and where the largest discrepancies occurred, especially around decisions the service expects staff to make. At the bedside, the patient who pauses halfway through a sentence still has something to contribute that the instrument cannot supply. Taking that observation seriously is entirely compatible with valuing pulse oximetry. Indeed, the best argument for examining the instrument so closely is how much we ask of it, from a quiet consultation to an urgent handover.

1On a follow-up visit after intensive care, a patient is offered a notebook. One entry describes the morning her daughter brought in a photograph. Another records the day she first sat out of bed. The dates are familiar to her family, who lived through them in sequence. For the patient, they belong to a fortnight she knows principally through a recurring dream about a railway station. A diary offers a way into that missing time. It is a considerable leap, however, from saying that it can answer a question about a lost fortnight to saying that it can prevent a psychiatric disorder.

2An intensive care diary usually has more than one author. A nurse can record why a procedure was necessary. A relative can describe the same afternoon by mentioning the radio programme that was playing. A medical chart could settle which procedure took place without explaining why a fragment of a song still troubles the patient. Conversely, a family account can bring a day to life while being decidedly vague about which day it was. The appeal of these entries lies partly in their different preoccupations. Editing them into a single seamless institutional voice might remove precisely the detail that makes the collection recognisable as one person’s story.

3An early European trial gave the practice a persuasive clinical argument. Christina Jones and colleagues arranged for one group of survivors to receive their diaries a month after leaving critical care. At three months, new cases of post-traumatic stress disorder were less frequent in that group than among controls: about five in a hundred compared with thirteen. For a unit deciding whether to devote scarce nursing time to a diary, the attraction of that finding is clear. Yet the calendar is a quiet participant in this result. The study concerned people who were available for enrolment a month into recovery and an intervention offered at that particular point.

4A later French trial examined diaries written by staff and families in 35 intensive care units. Of 657 patients randomised, 339 could be assessed three months after discharge from intensive care. About 30 per cent of assessed diary recipients, against 34 per cent of controls, crossed the study’s threshold for substantial post-traumatic stress symptoms. The difference did not reach statistical significance. The threshold was a score on a symptom questionnaire. It was possible for a patient to report troubling symptoms without a clinician having made a diagnosis. Describing this as a second count of newly diagnosed cases would make the two trials look more alike on paper than they were.

5It would be convenient to put the two studies to a vote and award the decision to the larger one. That shortcut overlooks who was still available to be counted. Random allocation makes the original groups comparable on average. It cannot guarantee that the people reached months later still represent everyone who entered. Missing follow-up might matter little, or it might matter a great deal if the likelihood of responding was linked to distress. The numbers alone cannot tell us which. Nor can uncertainty be recruited to either side as proof: it supplies neither a hidden benefit nor a guarantee that nobody was helped.

6Interviews with survivors open a different window. Some readers describe being able to ask a relative about an event that neither had previously found a way to mention. For others, reading can expose just how ill they were, at a point when they would prefer to be looking ahead. Set those accounts alongside an unchanged symptom score and there need be no contradiction. A conversation that is worth having can also be painful. Interviews illuminate what readers do with a diary and what opening it costs them. They do not establish how those same people would have fared if a diary had never been kept.

7Consider how a ward might judge its own service. Counting completed notebooks is simple, and a falling total could reasonably prompt a check on whether staff have time to write. Used as a measure of success, though, the count gives a completed book its full value before anyone has opened it. A patient who saves the notebook for a year and a patient who reads it with a nurse next week appear identical in the return. Adding a deadline for reading would solve that accounting problem only by handing the timetable to the service. The apparent improvement in the figures might come at the expense of the very choice the diary was meant to restore.

8A more defensible offer begins with a smaller promise. Explain who wrote the account, ask whether the patient would like to see it, and make room for a different answer at a later visit. A service interested in preventing post-traumatic stress should evaluate that claim with an outcome capable of testing it. A patient interested in finding out who brought the photograph has a different question, with an answer on a particular page. Both enquiries deserve to be taken seriously. Only one of them requires a trial, and neither gives the ward the right to decide when the cover should be opened.