Per-Seat Buckle Sensing with a Persistent Localised Audible Alert: A Systematic Review of the Passenger Car Evidence and Its Application to Buses

1Introduction

1.1Reported seatbelt-wearing rates

About 97 per cent of Australian car occupants wear seatbelts, compared with reported rates of 7 to 45 per cent among bus passengers.¹ One relevant difference is that car reminders can detect an unfastened belt, whereas the bus reminders discussed here do not.

The consequences are documented. In the rollover of coach 4666MO at Greta, New South Wales, on 11 June 2023, 6 of 35 passengers were wearing an available seatbelt. Ten passengers died. Fourteen were fully ejected and three partially ejected, and all ten deaths involved ejection. No belted passenger was thrown from their seat.¹ A Swedish review of 128 injuries across three coach rollovers also supported the protective value of seatbelts, identifying seatbelts as the single most effective countermeasure against ejection and injury severity.²

Wearing a fitted belt on a bus is already mandatory in New South Wales, but detecting and responding to non-use remains difficult. Across three financial years to 30 June 2025, one passenger fine was issued statewide.¹ Without buckle sensors, the bus cannot automatically detect and alert an unbelted passenger.

1.2Why this review draws mainly on car studies

The intervention this review examines is a per-seat buckle sensor paired with an alert that persists until the belt is latched. On buses, one related published trial was located, from Korea, and it used a light rather than a sound.³ On cars, the same intervention has been studied for more than two decades, in observational fleets, in randomised on-road experiments, in simulators and in national fatality data.

Car and bus systems can use the same basic process: detect an unfastened belt and produce an alert that stops when it is fastened. Buses differ in cabin size, sound conditions and the number of passengers who hear the alert. Section 6 examines how these differences may affect the results. Car evidence can inform bus design, but does not establish the effect on bus passengers.

1.3Questions

Four questions, answered in this order.

1. Does per-seat buckle sensing paired with an alert raise belt use in cars? Section 3.
2. Which design features contribute to the improvement? Section 4. The review considers alert duration, sound versus light, detection at each seat, false alerts and acceptability separately. These features help operators assess a system.
3. Why does it work? Section 5. Understanding why an alert works helps assess whether it could also work on a bus.
4. Could the same approach work on a bus? Section 6 examines each feature, including conditions that may reduce effectiveness.

Sections 7 and 8 discuss the direct bus evidence and regulatory requirements. Sections 9 and 10 set out design requirements based on the review.

2Method

2.1Search

Searches were run in PubMed, ScienceDirect, Google Scholar, the Insurance Institute for Highway Safety bibliography, the NHTSA publications catalogue, the Enhanced Safety of Vehicles conference proceedings and the European Commission road safety observatory. The searches covered English-language material published from 1990 to August 2026.

Search terms were combined from three groups: the intervention (seat belt reminder, safety belt reminder, buckle switch, buckle sensor, occupant detection, belt warning, interlock), the outcome (belt use, wearing rate, restraint use, compliance, fatality risk) and the setting (car, passenger vehicle, rear seat, bus, coach, motorcoach, school bus). Reference lists of every included study were then searched by hand, which is how the Swedish bus observation study and the Korean bus pilot were located.

2.2Inclusion and exclusion

A study was included if it reported a quantitative effect on observed belt use, measured belt use, self-reported likelihood of belt use, or a safety outcome such as fatality rate, and if the intervention involved detection of an individual belt’s state. Thirteen studies met these criteria and appear in Table 1 or Table 2. Regulatory instruments, standards and mechanism literature from outside road safety were included as supporting sources but are not counted as evidence of effect.

Studies were excluded if the intervention was information only with no sensing (these appear once, as a comparison set in Section 3.4), if the outcome was attitudinal with no behavioural or intention measure, or if the effect could not be separated from a concurrent enforcement or publicity campaign.

2.3How effect sizes are reported

Three kinds of number appear in this literature and they are not interchangeable.

• Percentage points are absolute changes. “Belt use rose from 71 to 76 per cent” is a gain of 5 percentage points.
• Per cent is a relative change. “Belt use increased 30 per cent” against a baseline of 40 per cent is a gain of 12 percentage points.
• Rate is the proportion wearing belts. “97 per cent wore belts” is neither a change nor a comparison.

Every figure in this paper is labelled. Effects are not pooled and no summary estimate is offered, for the reason given in Section 3.5.

3Does per-seat sensing with an alert raise belt use?

Yes, in every comparison located, across four study designs and four countries. Table 1 sets out the direct evidence. Section 4 examines design features that may contribute to these improvements. Differences between study populations and methods also affect the reported gains.

3.1Observational studies report higher belt use with reminders

The three large observational studies were run in different countries, on different fleets, by different teams, and all report higher belt use with reminders, although the initial wearing rates and comparison systems differ. Freedman et al. found 3 to 4 percentage points compared with an American reminder that already met the minimum chime requirement.⁶ Williams et al. found 5 percentage points against no reminder at a baseline of 71 per cent.⁴ Krafft et al. found 16.6 percentage points against no reminder at a baseline of 82.3 per cent.⁵

The Swedish study’s larger difference is consistent with the use of a more persistent reminder, although comparisons between studies cannot isolate the reason. The European Commission’s road safety observatory reports the same dataset as a three-way split: 82 per cent of drivers belted with no reminder, 93 per cent with a mild reminder, 99 per cent with an advanced one.¹⁷ Wearing rates were higher with more advanced reminders.

3.2Experiments compare reminder conditions in the same drivers

Observational studies compare vehicles, and vehicles differ in more ways than their reminder systems. Experiments address this problem by comparing the same drivers under different reminder conditions.

Kidd and Singer recruited 49 drivers who admitted to not always wearing a belt and who had a recent seatbelt citation, a group whose belt use is difficult to improve. Each drove a vehicle with three intermittent 7 second chimes, then a vehicle with either a 90 second audible reminder or an indefinite one. Belt use rose 30 per cent with the 90 second reminder and 34 per cent with the indefinite one, both relative to the intermittent condition.¹³ The authors put the national implication plainly:

“Strengthening existing U.S. safety standards to require audible reminders lasting at least 90 seconds for front-row occupants could save up to 1,489 lives annually.”

Kidd & Singer, Journal of Safety Research, 2019¹³

That result is the single most important finding in this review for present purposes, and Section 4.1 returns to why.

3.3Evidence on fatality reduction is less certain

Belt use is an indirect measure of safety benefit. The only study located that measures a safety outcome is Farmer and Wells, who compared driver fatality rates per registered vehicle for otherwise identical models with and without enhanced reminders, across United States data for 2000 to 2007. Fatality rates were 6 per cent lower with the reminder. After adjusting for differences in vehicle age, the manufacturer-specific estimates ranged from 9 per cent lower to 2 per cent higher.⁹

The differences between manufacturers are important when interpreting the result. The study supports the direction of effect. It does not support a precise fatality reduction figure.

3.4The comparison set: what happens without sensing

Signs and announcements provide a useful comparison because they give information without detecting belt use.

Nemire video-recorded rear seat passengers in rideshare vehicles in San Francisco and found that a warning sign at seated eye level produced significantly higher belt use than no sign.¹⁶ That is a real effect and it should be acknowledged: signage is not worthless. But the study reports a comparison of sign against nothing, in a four-seat cabin where the passenger has just chosen to sit down alone, and it produced no per-seat figure comparable to Table 1.

Australian crash evidence also shows the limitations of relying on signs and briefings. The Greta coach carried at least one sign promoting seatbelt use. Wearing was 17 per cent. The operator’s own policy required the driver to brief passengers verbally; the driver did not, and had not on other recent charters, and nothing in the system detected the omission. Enforcement resulted in one passenger fine statewide in three years.¹

3.5Why no pooled estimate is offered

The effect sizes in Table 1 cannot be combined into a reliable overall estimate for four reasons:

• Different baselines. Freedman’s comparator was an existing chime at a high baseline. Krafft’s was no reminder at 82.3 per cent. Park’s was no reminder at 9.6 per cent. A percentage-point gain means something different at each.
• Different populations. Kidd and Singer deliberately sampled cited part-time belt users. Krafft observed the general driving population. The first is a group with more scope to increase belt use.
• Different metrics. Table 1 contains observed belt use rates, relative changes measured in the same participants, rated intention in a simulator, and a fatality rate. These are four different quantities.
• Different units of randomisation, or none. Two studies randomised, the rest did not.

The studies generally support buckle-linked reminders, but do not provide a single effect estimate that applies across settings. Initial wearing rates, reminder design and study methods differ. Information-only measures have less consistent results. In particular, the 49.4 percentage-point difference in the Korean pilot should not be treated as the expected effect on other buses.

4Which design features contribute to effectiveness

Table 1 answers whether the intervention works. It does not tell an operator what to specify. Five features separate the systems, and the literature is unusually clear about three of them.

4.1Persistence: allow the alert to continue until the belt is fastened

This is the best-supported design finding in the review, and it comes from a direct experimental comparison rather than from comparing fleets.

The United States federal minimum, FMVSS 208, requires an audible signal of 4 to 8 seconds. Kidd and O’Malley report: that requirement does not increase seat belt use.⁸ Reminders exceeding it increase belt use by about 6 percentage points.⁸ Kidd and Singer then isolated duration as the variable, and found that reminders lasting at least 90 seconds beat three intermittent 7 second chimes by 30 to 34 per cent relative.¹³

Two qualifications. Kidd’s simulator study found that reported likelihood of buckling did not keep rising as duration increased beyond the enhanced threshold.¹¹ And the same study found that increasing the duty cycle, the proportion of each cycle the chime is actually sounding, raised annoyance without raising intention.¹¹ The findings support reminders lasting at least 90 seconds compared with the short chimes tested. They do not establish that every shorter reminder is ineffective, or that louder and more continuous sound is always better.

A reminder that continues until the belt is fastened can remain active for more than 90 seconds if needed. It stops earlier when the passenger buckles up: someone who fastens the belt after 4 seconds hears 4 seconds of alert.

4.2Alert type: sound compared with light alone

There is less evidence for alert type than for duration, but the car findings favour sound. Kidd, McCartt and Oesch found drivers more likely to buckle in response to a sound or vibration reminder than to a visual one.¹⁵ The effective car systems in Table 1 that used alerts rather than restricting vehicle operation used sound: BeltMinder chimes, the Euro NCAP systems chime, the Kidd and Singer conditions were audible by definition.

The mechanism is straightforward. A visual warning has to be looked at. A passenger may not look at a visual reminder. An audible warning can attract attention without requiring the passenger to look towards it, which is why sound is widely used for safety warnings.¹⁸

The single bus study in this review used a light only, and its authors say so. That matters enough to be treated separately, in Section 7.1.

4.3Per-seat detection, and the separate question of occupancy

Buckle and occupancy sensors provide different information and should be considered separately.

A buckle switch tells the system whether a belt is latched. Without it, the alert cannot respond to whether that belt is fastened. An occupancy sensor tells the system whether anyone is in the seat. Without it, the alert fires for empty seats.

Cars have fewer seats to monitor. Both UN Regulation No. 16 and NHTSA’s 2024 final rule permit a rear-seat design without occupancy sensing: rear seat systems are not required to detect occupancy, and instead alert only when a belt that was fastened is subsequently unfastened while the vehicle is moving.^19,20

This is an inference from the two regulatory texts and the arithmetic of bus occupancy, not a finding either text states. Park et al. reached the same design conclusion independently and fitted both sensors at every seat.³

4.4False alerts can cause passengers and drivers to ignore warnings

No study in Table 1 reports a false alarm rate. Evidence from other fields shows why this needs to be measured.

In clinical monitoring, the proportion of alarms that are false or clinically insignificant has been measured at 80 to 99 per cent, and the documented result is alarm fatigue: staff silence, disable or ignore alarms, including the ones that are correct.²¹ Repeated false alarms can reduce responses to true signals.

On a bus, repeated alerts at empty seats may lead drivers and passengers to ignore valid warnings. Increasing the volume or duration does not address the underlying sensor fault. Over a bus service life approaching 26 years,¹ false alerts may lead operators to disable the system. Checking that a seat is occupied prevents empty-seat alerts. Allowing the driver to disable only a faulty seat’s alert, as discussed in Section 6.2, keeps the remaining alerts operating.

4.5Acceptability, and the reason interlocks are excluded

A disabled system cannot prompt passengers to buckle up. Two studies examine acceptance and bypassing of the equipment.

Kidd et al. found a gearshift interlock raised belt use 16 per cent relative to an enhanced reminder, but 6 of 16 part-time belt users circumvented it.¹² Kidd and Singer found a speed-limiting interlock and a persistent audible reminder statistically indistinguishable in effect, at 33 and 30 to 34 per cent respectively, with more participants circumventing the interlock and reminders rated more acceptable.¹³

These findings suggest that persistent audible reminders can achieve similar improvements to interlocks while being more acceptable to users. A bus interlock also creates a practical problem: one unbuckled passenger could delay everyone.

5Mechanism

Understanding why reminders work helps assess whether they could also work on buses. Section 6 examines the relevant differences between settings. A similar mechanism does not guarantee the same effect size.

5.1Negative reinforcement: buckling up stops the alert

A persistent buckle-linked alert uses negative reinforcement: buckling up removes an unpleasant sound, encouraging that action. Behaviour analysis calls this an escape contingency. Van Houten et al. describe their gearshift delay in exactly these terms, noting that drivers “could escape or avoid the delay by fastening their seat belts”.¹⁰

This is why the intervention works on people that information campaigns do not reach. Buckling is the fastest way to stop the noise. Compliance does not require the passenger to believe a crash is likely, to have heard the announcement, to have read the sign, to remember a briefing, or to agree with any of it. The passenger can stop the alert by fastening the belt.

An announcement depends on several steps: the passenger must be present when it plays, hear it over the cabin, recognise it as addressed to them, understand it, accept it, and then act. A passenger may miss or not complete any of these steps. Kidd and O’Malley report that the 4 to 8 second federal chime does not increase belt use, although that finding does not identify which step failed.⁸

5.2The alert gives an immediate reason to buckle up

Leaving the belt unfastened increases the risk of injury if a crash occurs. Its immediate appeal is avoiding a belt that may be dirty, too short, awkward to reach past a neighbour, or uncomfortable to sleep in. Anund et al. document every one of those reasons from Swedish bus passengers.²² Passengers at Greta gave the same kinds of reason, including one who said a belt had stained another passenger’s clothing.¹

Passengers may give more weight to immediate discomfort than to the risk of a future crash. A persistent alert makes leaving the belt unfastened immediately unpleasant. It gives the passenger a reason to buckle up without needing to change their view of crash risk.

5.3Make clear which passenger needs to respond

Identifying the passenger who needs to respond is especially important in a large cabin.

A car chimes because it has detected that your specific belt is undone, and it goes silent when you fasten it. The warning responds to your unfastened belt. A bus with ADR 68/01 equipment plays a recording on a timer. It plays identically whether every belt is fastened or none are. It broadcasts the same message to everyone in the cabin.

The difference matters because individuals may feel less responsible for responding when a message addresses a group. Latané and Nida’s meta-analysis of 56 experiments found that in 48 of them, people were more likely to act when alone than when others were present.²³ The standard countermeasure in the applied literature is to address a specific person rather than a group. A cabin-wide announcement to 57 people is the group condition. A buzzer at one seat is the individual condition.

There is a second, separate reason localisation helps, from the warning design literature: directional auditory warnings produce faster and more accurate responses than non-directional ones in driving tasks.²⁴ That evidence comes from take-over requests in automated vehicles, not from seatbelts, and is offered as mechanism support rather than as an effect estimate.

6How the car findings may apply to buses

This section considers how detecting belt use, stopping the alert when the belt is fastened, and identifying the passenger who needs to respond may work on buses. It also considers practical differences between cars and buses.

6.1Low initial wearing rates allow more improvement, but do not predict it

Table 1 shows a clear pattern: the lower the baseline, the larger the absolute gain. Freedman’s 3 to 4 percentage-point gain was measured at a high initial wearing rate against an existing chime.⁶ Krafft’s 16.6 percentage-point gain was measured from an initial wearing rate of 82.3 per cent with no reminder at all.⁵ Park’s 49.4 percentage-point gain was measured from an initial wearing rate of 9.6 per cent.³

Baseline belt use constrains the possible gain: a population already at 97 per cent has three percentage points of room for improvement. Bus populations with reported wearing rates of 7 to 45 per cent have more scope to improve. This makes larger gains possible, but the car figures do not establish either a minimum or an expected bus effect.

6.2Bus conditions that may reduce effectiveness

Three conditions in Table 3 may make the system less effective or harder to operate on buses.

False alerts disturb passengers who cannot stop them. The person annoyed is often not the person who can stop it. In a car, the driver who hears the chime is the person whose belt is undone. In a bus, a false alert at seat 14 is heard by seats 12 through 16, none of whom can silence it. This may increase the risk that passengers and drivers ignore alerts, as discussed in Section 4.4.

A driver control can disable the alert at one faulty seat while leaving the other 56 operating. This reduces the need to switch off the whole system, a response to unreliable alarms documented in the alarm fatigue literature.²¹ This is a temporary measure. The affected seat has no alert until the fault is repaired, so operators need a process for recording and fixing it. Alerting only at occupied seats, a measured false alert rate and per-seat isolation are three separate requirements, and a bus system needs all three.

Vehicle interlocks are impractical for bus passengers. In cars, if reminders prove insufficient, an alternative is a gearshift or speed-limiting interlock, and the literature shows it works.^12,13 On a bus, an interlock could delay everyone because of one passenger. Enforcement is another option, but recorded enforcement has been rare.¹

The service life is two to three times longer. Every controlled study in Table 1 ran for weeks. The longest observational comparison spans a fleet, not a vehicle’s lifetime. The studies do not establish whether a passenger who has heard the same buzzer for fifteen years responds to it in year fifteen the way they did in year one. Section 12 treats this as the review’s principal limitation.

7The direct bus evidence

7.1The Korean pilot used visual alerts only

The Korea Automobile Testing and Research Institute fitted an interurban bus in commercial service with an occupancy sensor and a buckle switch at every passenger seat, a warning light at each seat, and a monitoring display at the driver’s position showing which seats were unfastened. Observed wearing was 9.6 per cent on the bus without the system and 59.0 per cent on the bus with it, a difference of 49.4 percentage points.³

The authors note an important limitation: the system carried no audible warning; passengers were reminded only by a light.³ They report the 40 to 50 point gain as achieved despite that limitation.

Set that against Section 4.2, where the car literature finds audible reminders more effective than visual ones,¹⁵ and against Section 4.1, where reminders lasting at least 90 seconds outperformed the short intermittent chimes tested.¹³ The bus pilot used visual alerts and reported a large improvement. It did not directly compare light with sound.

7.2The Swedish observation study makes the same recommendation independently

Anund, Forward and Sjörs Dahlman combined 328 bus observations across ten Swedish cities with seven focus groups and a survey of 1,737 respondents. Observed passenger wearing was 92 per cent in charter traffic, 50 per cent in commercial traffic and 27 per cent in regional traffic.²² Their focus groups with drivers found what the Greta investigation found in a different way, that a driver cannot see who is belted while driving.

Their recommendation list, written by a state-funded research institute with a declared absence of competing interests, contains this line:

“Ideally, the announcements could be replaced by seat belt reminders as used in passenger cars.”

Anund, Forward & Sjörs Dahlman, Accident Analysis and Prevention, 2023²²

That is the central recommendation of this review, reached independently from Swedish data by a national research institute, and published before ADR 68/01 was approved.

7.3Australian crash investigations: reminders and checks

The Greta investigation reviewed eight Australian rollovers of belt-fitted buses. Belt use was generally low in six of them. In two, Pentland Hills in 2022 and Exford in 2023, use was high, and the investigator attributed the difference to a single cause: “the reminders and checks by the driver and school staff”.¹

These cases support the value of reminders and checks, while showing that their delivery depends on drivers or staff carrying them out. They do not isolate the effect of reminders from other differences between the trips. At Greta, the required briefing was not given.

8What the rules require

8.1The comparison

What ADR 68/01 does require is worth stating fairly, because it is a genuine improvement on what preceded it: a label on the back of every seat, an illuminated cabin sign lit for at least 60 seconds after the last door closes, and one recorded announcement before the vehicle travels 500 m or exceeds 25 km/h. It also removed an exemption that had allowed some low-reference-height coaches to skip ADR 68 altogether.²⁶ Every one of those three measures falls in the comparison set of Section 3.4.

8.2Why the car cost-benefit case warrants assessment for buses

ADR 4/05’s seatbelt reminder requirement rests on a stated finding, quoted in the Regulation Impact Statement:

“the safety benefits of installing a driver’s seat belt reminder, over the life of a vehicle, outweigh the installation costs.”

Regulation Impact Statement, Vehicle Standard (ADR 4/05 Seatbelts)²⁵

The same three factors can be assessed for buses: safety benefit, vehicle service life and installation cost. The following points support examining that case.

1. Benefit. The reasoning was applied to a population already wearing belts at about 97 per cent and withheld from one wearing them at 7 to 45 per cent. Section 6.1 explains why lower initial rates allow more scope for improvement, without guaranteeing a larger effect.
2. Life of the vehicle. A car turns over faster than a coach, whose service life approaches 26 years.¹ A longer life spreads a one-off installation cost over more trips.
3. Cost per protected occupant. One car fitment protects one seating position. One coach fitment protects up to 57.

The reasoning behind ADR 4/05 has not been rebutted for heavy vehicles. It has not been extended to them. The Greta investigation draws the same distinction, recording that reminder systems were required in passenger vehicles while “similar systems were not required for coaches”.¹

8.3Rear-seat car reminders provide a passenger comparison

An obvious objection to Section 8.2 is that ADR 4/05 covers the driver, who controls the vehicle, and that driver findings may not apply directly to passengers. The 2024 United States final rule provides a relevant passenger comparison, because it extends per-seat sensing to people who do not drive.

From 1 September 2027, every rear seating position in a covered vehicle must have a seat belt warning system.²⁰ NHTSA estimates the rule will save about 50 lives and prevent more than 500 injuries a year, with benefits up to $831 million annually.²⁹ The population it protects is rear seat occupants, who in the United States wear belts at 80.4 per cent against 91.9 per cent in front.²⁷

Rear-seat car passengers and bus passengers share some relevant conditions. Neither is driving. Neither receives a warning aimed at them under current Australian rules. Neither can be reached by an interlock. The regulator that has looked hardest at this population concluded that per-seat sensing was justified for them at a baseline of 80.4 per cent. Australian bus passengers sit at 7 to 45 per cent, and their compliance date for a system with no sensor at all is 1 November 2027, seven weeks after the American rear seat rule takes effect.

There is a further difference between the requirements. NHTSA reports that the existing 4 to 8 second driver chime is the weakest form of the intervention and that enhanced reminders add roughly 3 to 4 percentage points over it.^6,14 Two regulators have now judged a 3 to 4 point improvement on a 92 per cent baseline worth mandating. Neither has been asked to consider a plausibly much larger improvement on a 7 to 45 per cent baseline.

9Design principles the evidence supports

These principles follow from Sections 3 to 6. They are written so that they can be applied to any manufacturer’s equipment.

1. Detect whether each belt is fastened. Buckle sensing allows the system to respond to an individual unfastened belt (3.4).
2. Alert only when the seat is occupied. This prevents alerts caused by empty seats and helps maintain trust in valid warnings (4.3, 4.4).
3. End the alert on latching, not on a timer. The buckle sensor tells the system when the passenger has acted (4.1).
4. Exceed roughly 90 seconds of potential duration. The experiment found reminders lasting at least 90 seconds more effective than the short intermittent chimes tested (4.1).
5. Use sound. Add light if you wish. A visual-only warning can be missed if the passenger does not look towards it (4.2).
6. Address the individual seat. An alert near the unbuckled seat may make clear which passenger needs to respond; this benefit still needs testing on buses (5.3).
7. Monitor the whole trip. Bus passengers unbuckle mid-journey to sleep, and a departure-time check cannot see it (6, Table 3).
8. Make buckling up stop the alert immediately. An immediate stop clearly links fastening the belt with ending the sound (5.1).
9. Measure the false alert rate and publish it. Set a threshold before deployment at which the system is judged to have failed, regardless of its effect on wearing rates (4.4).
10. Let the driver isolate one seat without disabling the system. The documented response to an unreliable alarm is to switch the whole thing off. A control for each seat lets the driver silence one faulty alert while keeping the other seats monitored (4.4, 6.2).
11. Do not assume louder or more continuous alerts work better. Higher duty cycle raised annoyance without raising intention (4.5, Table 2).

10An implementation assessed against the principles

The AITSC Seatbelt Warning System places two passive switch sensors at every seat, one in the cushion and one inside the buckle. Twenty seconds after the last door closes, any seat that is occupied but unbuckled triggers an 80 dB buzzer with an aisle-visible LED at that row, which continues until the belt is latched. Monitoring runs for the whole trip, so a passenger who unbuckles later is detected. The driver can disable the buzzer at an individual seat, so an alert at one faulty seat can be silenced while the rest of the system remains active. It is a separate product from the company’s ADR 68/01 compliance kit, which supplies the labels, illuminated sign, speakers and control module that the rule requires and which, by the analysis in Section 8.1, is an information-only measure.³⁰

10.1On the 80 dB level

The chosen 80 dB level aims to make the alert audible while limiting noise exposure.

Audibility. ISO 7731 treats a danger signal as clearly audible when it exceeds the effective masked threshold (the level needed to hear the signal over background noise) by at least 15 dB and is not below 65 dB(A), and recommends signal content in the 500 to 2,500 Hz band.¹⁸ Reported bus interior noise runs around 68 to 74 dB(A) in normal operation, with design targets near 65 dB(A) at maximum speed.³¹ The stated level is 15 dB above 65 dB(A). Whether it is sufficiently audible above background noise depends on operating conditions and measurement at the passenger’s ear.

In the condition where the buzzer usually fires, stationary with the doors closed and 20 seconds elapsed, cabin noise is well below the in-service figures above and the alert is expected to be more clearly audible. For a mid-trip unbuckling at highway speed the margin narrows to roughly 6 to 12 dB against the broadband cabin level. Two factors may improve audibility: ISO 7731’s margin is measured against the masked threshold in the signal’s own frequency band rather than against the overall A-weighted level, and the buzzer sits at the seat while cabin noise is diffuse, so the level at the intended listener’s ear is the design level rather than a level attenuated by distance. Neither of those is a substitute for in-cabin measurement, and the mid-trip case is where a deployment should be verified rather than assumed.

Exposure. Sound level must also be considered alongside exposure duration. Occupational exposure standards treat 85 dB(A) as the level at which sustained exposure requires control, and an 80 dB alert sits below it. A louder device could be easier to hear at speed but would also increase passengers’ noise exposure. The exposure question is otherwise governed by duration, and a correct alert is terminated by a one second action. A false one is not, because a passenger cannot silence a buzzer triggered by a fault in their own seat. The driver therefore needs a way to silence a faulty seat alert. Sound level, false-alert frequency and the ability to stop a faulty alert must be assessed together.

11What is still not known

Two questions this review cannot answer from existing literature.

Are audible alerts more effective than lights on a bus? Park et al. tested a light and got 49.4 percentage points.³ The car literature says sound should do better.¹⁵ Nobody has run the comparison in a bus.

Is an alert near the seat more effective than the same sound played throughout the cabin? Section 5.3 gives the mechanism and Section 6 gives the reasoning, but no study in this review varied the location of the alert. This is a central design hypothesis that has not yet been tested.

Both are answerable by one trial: a randomised controlled trial with buckle and occupancy sensors fitted to every vehicle in every arm including the control, so the control group is measured with the same instrument as the treatment group, and vehicles randomised to no alert, a light at the seat, the same sound played cabin-wide, or a localised persistent buzzer. The primary outcome is the proportion of occupied seats latched five minutes after the last door closes. False alerts per 1,000 occupied seat-trips is a secondary outcome with a pre-registered failure threshold, and driver isolation events are logged alongside it, because disabling faulty seats without repairing them can reduce the recorded false-alert rate while leaving fewer passengers monitored. A full protocol, including power calculations and the failure conditions published in advance, is set out separately.³²

12Limitations

Three main limitations follow, in order of importance.

1. Long-term response to repeated alerts has not been measured. Every controlled study located ran for weeks. Passengers may remain more responsive to alerts they can stop by buckling up than to reminders that play regardless of their actions. This hypothesis has not been tested over a vehicle’s service life.
2. A buckle switch measures latching, not restraint. It cannot see a belt sat on, worn behind the back, or a neighbouring belt clipped into the wrong buckle. Sensor records may count belts that are latched but not worn correctly. Direct observation is needed to measure how often this occurs.
3. Australian bus wearing-rate data is limited. The 7 to 45 per cent range comes partly from crash samples rather than systematic roadside observation, and passengers involved in crashes may not represent passengers on ordinary trips.

No single measure would have prevented every outcome at Greta. The investigation identified glazing, seat design, the road environment and driver health among the contributing factors, and recommended anti-ejection glazing alongside the seatbelt recommendations.¹ Raising wearing rates addresses one factor among several.

13Conclusion

Car studies support buckle-linked audible reminders, particularly persistent reminders compared with short chimes. The studies include observed belt use, reported intentions and fatality outcomes, which must be interpreted separately. The evidence supports detecting buckle state, allowing the alert to continue and stopping it when the belt is fastened.

The proposed mechanism is negative reinforcement: fastening the belt stops the alert. The review examines how that mechanism can apply to bus passengers. Sensors can link each seat’s belt state to an alert. In a large bus cabin, placing the alert near the unbuckled seat may clarify who needs to act. The benefit of that arrangement is a hypothesis supported by behavioural reasoning, not a measured bus result.

The difference between car and bus reminder requirements warrants further assessment. Per-seat sensing was mandated for Australian car drivers wearing belts at 97 per cent, and is being mandated for American rear seat passengers wearing them at 80 per cent, using cost-benefit reasoning that could also be assessed for buses. It has been withheld from Australian bus passengers wearing them at 7 to 45 per cent, whose new rule takes effect in the same year as the American one and requires no sensor at all. The reasoning behind the car rule has never been rebutted for buses. It has simply not been extended to them.

References

1. Office of Transport Safety Investigations (NSW). Bus safety investigation report: Rollover of coach 4666MO, Greta, NSW, 11 June 2023. Sydney: OTSI.
2. Albertsson P, Falkmer T, Kirk A, Mayrhofer E, Björnstig U. Case study: 128 injured in rollover coach crashes in Sweden, injury outcome, mechanisms and possible effects of seat belts. Safety Science. 2006;44(2):87–109.
3. Park et al. The effectiveness of seat belt reminder (SBR)s by vehicle types. Proceedings of the 27th International Technical Conference on the Enhanced Safety of Vehicles (ESV), Paper 23-0190. Korea Automobile Testing and Research Institute; 2023.
4. Williams AF, Wells JK, Farmer CM. Effectiveness of Ford’s belt reminder system in increasing seat belt use. Injury Prevention. 2002;8(4):293–296.
5. Krafft M, Kullgren A, Lie A, Tingvall C. The use of seat belts in cars with smart seat belt reminders, results of an observational study. Traffic Injury Prevention. 2006;7(2):125–129.
6. Freedman M, Levi S, Zador P, Lopdell J, Bergeron E. The effectiveness of enhanced seat belt reminder systems: observational field data collection methodology and findings. DOT HS 810 844. Washington DC: NHTSA; December 2007.
7. Lie A, Krafft M, Kullgren A, Tingvall C. Intelligent seat belt reminders, do they change driver seat belt use in Europe? Traffic Injury Prevention. 2008;9(5):446–449.
8. Kidd DG, O’Malley S. Increasing seat belt use in the United States by promoting and requiring more effective seat belt reminder systems. Traffic Injury Prevention. 2023;24(sup1):S80–S87.
9. Farmer CM, Wells JK. Effect of enhanced seat belt reminders on driver fatality risk. Journal of Safety Research. 2010;41(1):53–57.
10. Van Houten R, Malenfant JEL, Reagan I, Sifrit K, Compton R. Increasing seat belt use in service vehicle drivers with a gearshift delay. Journal of Applied Behavior Analysis. 2010;43(3):369–380.
11. Kidd DG. Response of part-time belt users to enhanced seat belt reminder systems of different duty cycles and duration. Transportation Research Part F: Traffic Psychology and Behaviour. 2012;15(5):525–534.
12. Kidd DG, Singer J, Huey R, Kerfoot L. The effect of a gearshift interlock on seat belt use by drivers who do not always use a belt and its acceptance among those who do. Journal of Safety Research. 2018.
13. Kidd DG, Singer J. The effects of persistent audible seat belt reminders and a speed-limiting interlock on the seat belt use of drivers who do not always use a seat belt. Journal of Safety Research. November 2019.
14. National Highway Traffic Safety Administration. Effectiveness and acceptance of enhanced seat belt reminder systems: characteristics of optimal reminder systems. DOT HS 811 097. Washington DC: NHTSA; February 2009.
15. Kidd DG, McCartt AT, Oesch NJ. Attitudes toward seat belt use and in-vehicle technologies for encouraging belt use. Traffic Injury Prevention. 2014;15(1):10–17.
16. Nemire K. Warning signs to fasten seat belts result in higher rates of rear seat belt use in rideshare vehicles. Proceedings of the Human Factors and Ergonomics Society Annual Meeting. 2019;63:2046–2050.
17. European Commission, European Road Safety Observatory. Seat belt reminders. eSafety statistics and analysis archive.
18. International Organization for Standardization. ISO 7731:2003, Ergonomics: danger signals for public and work areas, auditory danger signals. Geneva: ISO; 2003.
19. United Nations Economic Commission for Europe. UN Regulation No. 16, safety belts, restraint systems, child restraint systems and ISOFIX child restraint systems, 08 series of amendments, including Annex 8 safety-belt reminder requirements.
20. National Highway Traffic Safety Administration. Federal Motor Vehicle Safety Standards; Occupant Crash Protection, Seat Belt Reminder Systems, Controls and Displays. Final rule, 49 CFR Part 571, Docket No. NHTSA-2024-0071. Federal Register 90 FR, 3 January 2025. Effective 4 March 2025; compliance 1 September 2026 (front) and 1 September 2027 (rear).
21. Cvach M. Monitor alarm fatigue: an integrative review. Biomedical Instrumentation & Technology. 2012;46(4):268–277.
22. Anund A, Forward S, Sjörs Dahlman A. Seat belt usage in buses, an observation study of usage and travellers’ perspectives. Accident Analysis and Prevention. 2023;190:107138.
23. Latané B, Nida S. Ten years of research on group size and helping. Psychological Bulletin. 1981;89(2):308–324.
24. Arabian A, Goodridge CM, Merat N. Asynchrony of directional auditory warnings and visual information in hands-on SAE level 2 automated vehicles: analysis of driver responses. PLOS ONE. 2025;20(11):e0336941.
25. Department of Infrastructure and Transport (Australia). Amendments to Vehicle Standard (Australian Design Rule 4/05, Seatbelts) 2006: Regulation Impact Statement. Office of Impact Analysis. See also Vehicle Standard (Australian Design Rule 4/05, Seatbelts) 2012, F2012L00468.
26. Vehicle Standard (Australian Design Rule 68/01, Occupant protection in buses). Approved 9 December 2024. Applies to new bus models from 1 November 2026 and all new buses from 1 November 2027.
27. National Highway Traffic Safety Administration. Occupant restraint use in 2023: results from the NOPUS controlled intersection study. Washington DC: NHTSA.
28. Regulation (EU) 2019/2144 of the European Parliament and of the Council of 27 November 2019 on type-approval requirements for motor vehicles as regards their general safety and the protection of vehicle occupants and vulnerable road users.
29. National Highway Traffic Safety Administration. NHTSA finalizes seat belt reminder rule to increase seat belt use and improve occupant protection. Press release, December 2024.
30. AITSC. Seatbelt Warning System and ADR 68/01 Kit: technical summary. 2026.
31. Interior noise measurements for buses. Altoona Bus Research and Testing Center, interior noise and vibration test reports (73.6 to 74.6 dB(A) at passenger seats during acceleration, 74.3 dB(A) average); and Lee YJ, A study of cabin inside noise of high-speed passenger coaches, WIT Transactions on the Built Environment (design target below 65 dB(A) at maximum speed).
32. AITSC Research Division. A randomised controlled trial of per-seat buckle sensing with a persistent localised audible alert on buses: protocol. 2026.