One hundred and twenty electric care beds were delivered with the top of the height range exactly right and the bottom 70 mm too high, and one substituted actuator explains the whole of it.

The order, and the number that came out of a care home

An aged-care provider in Europe was replacing the beds in two of its homes and ordered one hundred and twenty electric care beds from a factory in Guangdong. Ninety-six for a residential wing, twenty-four for a smaller unit for residents with reduced mobility. Each bed had a four-section platform, a powered backrest and knee rest, height adjustment on a vertical lift actuator at the centre of the frame, and a 100 mm pressure-relief mattress supplied by the buyer.

The purchase order was ordinary in every respect except one line. It named the mattress surface height as a range, 350 mm at the low end and 780 mm at the top, and the range was right. At the low end the platform has to come down close to the floor so that a resident sitting on the edge can put both feet flat and push up. At the top it has to be high enough for a carer to work over the bed without stooping.

The number did not come from a supplier's catalogue. It came from the group's own occupational therapist, who had measured the bed the group already owned: at its lowest setting, with a resident on the edge, the top of the mattress sat 350 mm above the floor, and the shortest resident in the home could get both feet flat on the ground and stand. The old beds had a lever under the frame and a release catch. The new beds were electric, and the whole point of the new beds was to do that job without anyone crouching beside the frame with a spanner.

The buyer had an approved sample. It was the unit the group's operations manager had seen at the factory the previous year, on the previous order, and it reached 350 mm. It was still in the factory's showroom when the inspector arrived for the new batch.

How the height was measured, and what had to be measured

The inspector worked from the purchase order's dimension list and the factory's production drawing, and the first thing he did was measure the bed in the state it would be used in, with the 100 mm mattress the buyer had specified, not on a workshop test rig with nothing on it.

That distinction decides the whole case. The drawing gave platform heights with the mattress off. The order gave mattress surface heights with the specified mattress on. The two differ by exactly the thickness of the mattress, so both readings were taken on every bed: platform, mattress removed, and surface, with the mattress fitted. The platform was read at the seat section, in the middle of the frame, because that is where a resident's feet are.

Each position was recorded twice, bed empty and bed loaded, and the load was the number on the drawing: 120 kg distributed across the platform in the pattern a seated resident's weight makes, rather than a person lying down on it.

Heights were read with the backrest and knee rest flat, at both ends of the travel, on a level floor, with the castors that were actually fitted to the bed. A bed measured with the backrest up is a different number, and a bed measured on a factory floor with one wheel on a ramp is a different number again.

The actuator had the same reach and a different folded length

A vertical lift actuator is described by two numbers, and on a furniture drawing only one of them usually appears. There is the stroke, the distance it pushes out, and there is the retracted length, the height it takes up when it is folded. The platform's lowest position is the base frame plus the retracted length. Its highest position is the base frame plus the retracted length plus the stroke. The distance the platform actually travels is the stroke on its own.

Those two numbers can be traded against each other. An actuator with a retracted length of 70 mm and a stroke of 430 mm is 500 mm long when it is fully extended. An actuator with a retracted length of 140 mm and a stroke of 360 mm is also 500 mm long when it is fully extended, fits the same space under the platform, carries the same load rating, and pushes the platform to exactly the same height at the top of the travel. It simply starts 70 mm higher and stops 70 mm lower.

That is what the fitted actuator did. The bed reached the top of its range at exactly the specified height, because the top of the range is set by the fully extended length, and the extended length had not changed. The bottom of the range was 70 mm too high, because the bottom is set by the folded length, and that had.

The inspector found it by putting a tape on the frame with the mattress off and reading 320 mm where the drawing said 250 mm, then asking the factory for the actuator's own data sheet. The label on the actuator was legible, and it was not the part on the drawing.

The approved sample and the batch were not the same bed

The buyer's file held an approved sample: a unit built in the previous production run, in the previous year, with the actuator the drawing called for. Comparing the batch against that sample was the fastest way to establish what had changed, and it was also the reason the fault survived. The substituted actuator is mounted under the platform, where a visual comparison of one bed against another will never look at it.

The label on the actuator gave the answer. Same extended length, same load rating, a different model number and a different retracted length. The factory's batch records carried a note from production planning: drawing part on six-week lead time, in-stock equivalent fitted, extended length and safe load unchanged. No approval had been sought from the buyer, because nothing in the order said the actuator was a fixed part.

Nobody at the factory set out to build a different bed, and nobody was trying to hide anything. The substituted part was not cheaper and did not carry less. It was the same price and the same rating, and on a drawing where it appeared as one model number in a parts list, it looked like the part it replaced.

What 70 mm does to a resident sitting on the edge of the bed

The group's care manual sets out what a resident has to be able to do from the lowest position of the bed: sit on the edge with both feet flat on the floor, lean forward, and stand. The height that makes this possible is set by the resident's lower leg, not by the bed. An adult's lower leg is roughly a quarter of their height, so a resident of 1.50 m has about 370 mm between the back of the knee and the floor, and a resident of 1.70 m has about 420 mm.

At a mattress surface height of 350 mm, both of those residents can sit with their feet flat and their knees flexed, and push up. At 420 mm, the feet of the shorter residents hang clear of the floor with nothing to push against. The taller residents can just about find the floor with the balls of their feet, which is not the same thing as standing up from it.

The consequence is not abstract. Every transfer that was going to be done by the resident becomes a transfer done by two staff members with a hoist, several times a day, for the rest of that resident's life. Staff handling is the largest single source of shoulder and back injury in care homes. It is also the difference between a resident who keeps getting up and walks to the dining room, and one who stops trying and spends the day in bed.

There is a second effect a drawing-based check misses. A hoist sling has to be fitted and tensioned around a resident sitting on the edge of the bed, and the sling's clearances behave differently at different bed heights. A bed sitting 70 mm too high moves the resident's weight up and away from the carer's best lifting position, which is why the care manual treats the low position as a staff safety requirement as much as a resident one.

The group's own measurement of their old bed was 350 mm. The sample the factory showed them was 350 mm. The specification said 350 mm. The only thing in the whole chain that was not 350 mm was the batch the inspector was standing next to.

How the finding was written up

The report records what was measured, the conditions it was measured under, and the difference from the specification, in numbers. The dimensions section of a report carries the specified value, the measured value, the tolerance and the measurement condition. It does not carry the phrase within tolerance, because that phrase is a conclusion and a conclusion needs the numbers behind it.

The dimensional finding was graded Major, seventy millimetres outside the stated tolerance. Under the buyer's own purchasing terms the substitution of a named component without approval is a Critical finding, and it was recorded as one, because it is what caused the dimensional finding and because it would have happened again on the next order. The recommendation was to hold the lot.

The report also states what could not be verified, which on this order was a short list but not an empty one: the actuators' own cycle life rating was not tested; the factory's incoming inspection record for the substituted part was not witnessed; and whether 350 mm suits this group's residents is the occupational therapist's judgement, not something an inspector can settle on a factory floor. That last one belongs to the buyer, and the report says so rather than implying the inspector had checked it.

The lot was one hundred and twenty beds. Under ISO 2859-1 at general inspection level II a lot of that size is sample code F, which gives twenty units, with an accept number of one and a reject number of two for major defects at AQL 2.5. All twenty failed. The buyer's purchasing terms required the lot size, sample size and accept and reject numbers to be stated, so they were stated, and the recommendation followed from the finding rather than from the count.

Twenty out of twenty is not the same as a rate of one hundred per cent

A finding of twenty defects in twenty sampled units is not a defect rate of one hundred per cent, and a report that says so is more useful than one that does not. The distinction matters when the buyer decides what to do next.

This defect was not random. It was a property of the build. Every bed in the lot carried the same substituted actuator, and every one of them sat 70 mm too high at the low end of the travel. So the inspector documented the finding on three further beds as well as on the twenty, put the actuator model number against the drawing, and reported the extent as the whole lot, with the reasoning written out rather than implied.

The report also says what the sample did not do. If the fault had been random, twenty units would have told the buyer something about the odds of it appearing in the rest of the lot. It did not, and the report says that the sample documents the finding rather than measuring it. The extent is a fact about the build, established by reading a model number off a label, not by counting failures.

That is why the recommendation was to fit one component to all one hundred and twenty beds rather than to sort the lot into good and bad units. Sorting is the right answer for a random defect. For a systematic one it is the wrong answer, and on this order it would have been an expensive one.

What the specification should have said

Nothing in this order was exotic. The fix is entirely in the purchase specification, and it is the kind of thing that belongs in a purchase order the first time and in a standing specification thereafter.

• State the datum and the condition with the number. Mattress surface height at the lowest position, minimum 350 mm above the finished floor, measured with the 100 mm mattress specified in clause 4 fitted, backrest and knee rest flat, on a level floor, with the castors as supplied. A height without a datum is a preference, not a requirement.
• Give the heights as named positions with a purpose, not as a bare range. Low and transfer position, mattress surface 350 mm maximum, so a resident can stand and transfer unaided. Working position, mattress surface 700 mm plus or minus 20 mm, for bedside care with the carer seated. High and bed-making position, mattress surface 780 mm minimum, for bed making and for a second carer working from the far side.
• Require travel, and require it loaded. Travel between the low and high positions shall be not less than 430 mm, measured under a 120 kg load distributed across the platform in the pattern of a seated resident, and the platform shall still reach the low position with the mattress fitted and the load applied.
• Say where deflection is measured and set a limit for it. Platform deflection under a 120 kg distributed load shall not exceed 8 mm at the mid-span of the seat section. The middle of the bed is where a resident sits and where the platform is weakest, and no drawing on this order had an opinion about it.
• Control the parts that set the dimensions. The lift actuator make, model, retracted length and stroke are part of this specification, and substitution requires written approval before production. Two numbers on a label, and this entire failure mode disappears.
• Approve a first article built from production materials, not a showroom unit. One bed from the production batch, measured at the factory with the mattress fitted and accepted in writing before the balance of the batch is built. The showroom unit was the trap here: it was the right bed, built the year before, with the part that was in stock then.

The part that was still in the warehouse

The substituted actuator was, at the time of the inspection, still the part the factory was building that bed with, because it was the one on the shelf. The fix on this order was to fit the specified actuator to all one hundred and twenty beds and re-inspect. The fix that would have mattered more was to take the actuator out of the parts list and put it in the specification, where a buyer can see it, approve it, and hold the factory to it.

The same defect has a different name in every category, and it is the same defect. The bracket that holds a sofa rail 30 mm lower than the drawing. The slide on a drawer that is a different model. The veneer on a table that is thinner than the claim. In each case the goods are close enough to right that a photograph shows nothing and a carton count shows nothing, and the difference appears only when somebody measures the specific dimension the buyer cares about, under the condition the buyer actually uses it in.

Which is the argument for an elderly care furniture inspection in one sentence. The value of an inspection is not the count of defects it finds. It is the number of dimensions in your specification that were quietly undefined.

What the inspection was worth

This was two days of an inspector's time at the factory: measure, compare against the drawing and against the approved sample, and write it up. It sits at the Full Verification level of the service, $299 per man-day, because the work included verifying a component against the drawing and against the buyer's sample rather than only counting units and looking at them.

Against that sits one hundred and twenty beds that could not be loaded, a container that had to be held, and an actuator substitution that would have repeated on every order the group placed with that factory until somebody wrote the part number into the specification. A finding that is on the report is worth arguing about. A finding that is not on the report is a bed that arrives at 420 mm and stays there for the life of the resident.