How the Home Safe Index is scored

What this is, and what it is not

This is a structured checklist for a home. It is not a medical or clinical assessment. It cannot diagnose anything, it does not evaluate a person’s health or mobility, and it does not replace an occupational therapist walking through the home in person.

We are not aware of any comparable free tool that publishes its weights at all. That does not make ours correct. It does make them checkable.

The shape of the assessment

The assessment covers five areas. Three are the core, asked of everyone. Two are optional and offered afterwards.

A few of those questions are routing questions that produce no score. They exist to skip things that do not apply, so someone with a walk-in shower is never asked about tub walls.

Nobody sees all 56. Depending on the home and the person, a typical assessment shows between 22 and 45 questions.

If you decline the optional areas, the remaining three are renormalized so they still sum to one, keeping their relative proportions. Declining does not penalize your score, and it does not inflate it.

How a single question is scored

Each question is a control that is either present or absent. When the control is absent, the question contributes risk. When it is present, it contributes nothing.

The contribution is:

Hazard weight runs from 1 to 10 and describes how serious it is for that control to be missing. Across the whole bank the weights in use run from 4 to 9. Nothing is a 10, because we did not find a single home condition we were confident enough about to call maximal.

Exposure multiplier adjusts for the person. Each factor applies only if the question is tagged as relevant to it:

These multiply together and are then capped at 2.5. Without the cap, someone with four risk factors would score more than four times a person with none, which overstates what we actually know. The cap is a deliberate limit on our own confidence.

Frequency multiplier adjusts for how often the hazard is met: 0.6 for seasonal, 0.9 for daily, 1.1 for several times a day, 1.3 for several times a day including at night. The same missing grab bar in a guest bathroom and in the only bathroom are not the same finding.

An answer of “not sure” scores at 70% of the full value. Uncertainty is not the same as safety, and it is not the same as danger.

Some questions are reverse-scored, where a “yes” is the risk. “Do they hold onto the towel bar to steady themselves?” is the clearest example, and it is one of the most informative answers in the whole assessment, because it identifies where support is needed in the person’s own behavior rather than our inference.

From questions to a score

Each area’s score is the sum of the risk actually present, divided by the maximum risk that area could have produced for this particular home, expressed as 0 to 100. Skipped questions are removed from both halves of that fraction, so a home with no stairs is not rewarded or punished for the questions it never saw.

Scores are reported as bands, not numbers:

We show you the band and not the number on purpose. The precision of a two-digit score is not precision we have earned, and a family arguing about whether 61 is meaningfully worse than 57 is having the wrong conversation.

The overall score, and why it is not an average

The overall figure is:

Half the score is the single worst area in the home. This is deliberate. A spotless kitchen does not make a staircase with no handrail safer, and averaging would let five decent rooms hide one dangerous one. Falls do not average.

Severity ceilings. The five areas are not equally capable of making a home dangerous, so each area has a limit on how far it can push the overall band through that “worst area” term:

• Bathroom, stairs, and night path can reach critical
• Living areas can reach high
• Kitchen can reach moderate

A kitchen where every single answer is bad is a real problem worth fixing. It is not the same problem as a staircase with no handrail, and the score should not say it is.

The single-finding override. If any one question in the bathroom, stairs, or night path produces a risk value above 15, the overall band cannot be reported better than high, no matter how good the rest of the home is. When this happens we name the finding that caused it, rather than quietly moving the number. Some things are serious on their own.

Two scores, not one

We report likelihood and consequence separately and never blend them.

Likelihood is how likely a fall is, given the home and the person.

Consequence is how bad it would be if one happened. Whether a phone can be reached from the floor. Whether the bathroom door could be opened from outside if someone fell against it. Whether anyone would know.

These are different problems with different fixes. A parent who is steady on their feet but alone nine hours a day has a low likelihood and a high consequence, and every tool we looked at collapses that into one number that describes neither situation. If fewer than three consequence questions apply to a home, we show the findings as a list and no band at all, because a score built on two questions moves 50 points on a single answer and that is noise wearing a number.

How the recommendations are ordered

The list is sorted by how much risk a fix removes relative to how much trouble it is:

Effort runs 1 to 5 and is about difficulty in the real sense, not labor. Effort 1 is turning down a water heater. Effort 3 is anything requiring an ongoing habit change or a conversation with someone who does not think they need help, which is where most of these actually fail. Effort 5 is moving a bedroom downstairs.

Permission penalty adds 3 if you rent and the fix needs landlord approval. Renters get told about those fixes, they just get sorted below things that can be done today.

Fixes are then grouped into this weekend, this month, and get a professional, using both cost and effort. This is why “relocate the bedroom to the main floor” never appears under “this weekend” even though it costs nothing.

Free fixes always display first. 52 of the 54 scoring questions have a remediation that costs nothing, and where a no-cost fix exists it is shown above every paid option for that finding, regardless of where the scoring math ranked it. Removing a rug beats buying a rug gripper. We would rather tell you that than sell you the gripper.

Findings we deliberately promote out of order

Three answers override the sorting entirely, because the ordinary math undersells them:

Products we will not recommend

Ten findings route to a referral instead of a product, and several product categories are blocked outright in code, so they cannot be recommended even by mistake:

• Adult portable bed rails and bedside assist bars. These are subject to a mandatory CPSC standard and have been associated with hundreds of entrapment deaths. Non-compliant units have been sold on major retail sites recently. We do not link them at any price.
• Suction-mounted and clamp-mounted grab bars. They fail on textured and grouted surfaces, and the failure happens in a bathtub.
• Step stools used to reach or to enter a bathtub. These remove one hazard by adding another. Where our assessment finds someone climbing, the recommendation is to remove the reason to climb.
• Canes, walkers, rollators, and wheelchairs. Height and fit matter, and the wrong one causes falls. These need a person, not a link.
• Subscription medical alert devices, hearing devices, diagnostic devices, supplements, and anything prescription or normally billed as durable medical equipment.

How this makes money, and why it does not affect your score

We earn a commission if you buy something through a link on the results page, at no additional cost to you.

The scoring runs to completion before any product is looked up. Weights are attached to questions, not to products. There is no product in our catalogue that can change a score, and where a fix costs nothing that fix is displayed above the paid one every time.

Cost figures shown next to a fix are our own estimates for that kind of work. They are not current prices from any retailer.

What we would fix about this

Being honest about the gaps is part of publishing the method.

If you use this professionally and think a weight is wrong, we would like to hear it. That is the point of publishing them.