Deficient hand washing facilities in public toilets in the time of the COVID-19 pandemic: A survey in one high-income country

Survey sampling

The sample frame was comprehensive for the public toilets in three contiguous cities with a combined population of 389,000, with additional convenience sampling in cities and rural areas, based on author travel plans for other reasons around New Zealand from 18 July 2020 to 2 January 2021. We expanded the sample to a total of 400 toilets, which we estimated would be at least 10% of the facilities nationwide. But we also aimed to replicate a previous survey of facilities in public toilets in the lower North Island in 2012/2013.¹¹ The public toilets were located by:

• Online city council maps of public toilets were used for the comprehensive three council area survey.^18-20 However, data could not be collected on a minority of these (<5%) due to closure for repairs and/or being padlocked shut, suggestive that they were only opened for specific sporting events.

• Using ‘Google Maps’ to locate a ‘city/town/district’, and then in the map searching for ‘public toilets’.

• Using the smartphone app version of ‘CamperMate’.²¹

• Roadside public toilet signs when travelling by car.

We excluded from the sample, public toilets that were not directly open to the outside (ie, which were inside of other buildings such as within shopping complexes; council-owned facilities such as libraries; and within railway or ferry buildings, some of which were signed as for “patrons only”). Where facilities were closed or were being cleaned, we attempted repeat visits where this was convenient.

Data collection

At each toilet facility, we surveyed all the men’s and unisex toilets. Toilet facilities contained from one to six self-contained “toilet units” with a separate door that opened to the outside – albeit some units of “men’s toilets” would include multiple hand washing basins, a urinal and multiple toilet bowls. Data were collected on the availability of water and soap for handwashing. Toilets with empty soap containers were counted as without soap. We also photographed all health-related signage (eg, relating to COVID-19 and handwashing). Requirements for touching (or not) of taps/buttons/levers for activating tap water and flushing the toilet were noted. The presence of lids for toilet bowls was documented, given concerns around potential virus dispersion when flushing when the lid is not down (see Introduction).

Analysis

We compared results for the three city council comprehensive sample with the supplementary convenience sample. Comparisons were also made with the exact same facilities involved in the previous survey conducted in 2012/2013 (albeit excluding 14.7% (22/150) which were not readily accessible or which had subsequently been closed down). Statistical analysis used OpenEpi (v 3.01) and Mid-P exact values were calculated (2-tailed).

National denominator estimation

To provide context for our survey, we estimated the total number of such public toilet facilities in New Zealand using the following steps:

• For city-based local authorities we used as a basis the facilities we surveyed in the three city comprehensive sample (n=131 or 3.4 per 10,000 population).

• For the other local authorities we used data from Google Maps in three LAs in the Wairarapa region (n=13 facilities) and scaled from our estimate of the sensitivity of such data on Google Maps from the three city council survey (at 61.8% or 81/131 giving a scaling factor of 1.62). This gave 21 facilities or 4.3 per 10,000 population.

• We then extrapolated the three city council results to the other 12 LAs in New Zealand which were city ones; and the Wairarapa results to the other 51 LAs which were council districts.

Survey results for 2020/21

Our survey comprised 400 toilet units at 242 toilet facilities. We estimated that nationwide the total number of such toilet facilities was around 985 in city councils and around 760 in district councils, ie, around 1864 in the whole country (3.7 per 10,000 population). So our sample was estimated at 13.0% of the estimated total (242/1864), which was greater than our target of a 10% sample.

Most of the sample of 400 toilet units were from the comprehensive survey of three contiguous city councils (n = 228), relative to the additional convenience sample (n = 172). The former grouping was all in the lower North Island, were more likely to be unisex (vs men’s toilets), and involved more toilets in cities (but fewer in towns and rural areas) (Table 1).

The proportion of all the 400 toilets without water for hand washing was 2.5%, with this being higher in the convenience sample than the three urban councils sample (5.2% vs 0.4%; risk ratio (RR) = 11.9; 95%CI = 1.53 to 93.3; p = 0.0030). Absence of soap was 14.8% overall, and this was also higher in the convenience sample vs the three councils sample (21.5% vs 9.6%; RR = 2.23; 95%CI = 1.37 to 3.64; p = 0.0011). Most soap was dispensed as a liquid or foam (Figure S1-3), but at 6.0% of toilets it was available in a cake/bar form. A number of toilets had containers for liquid soap, but they were empty (Table 1).

In terms of “no touch” activation, this was available for hand washing water at 28.0%, for toilet bowl flushing at 18.5%, and for urinal flushing at 80.5%. Toilet bowl lids were not present for 32.8% (many were designed or built this way) (Figure S3-1), and 2.3% of toilets had damage which would impair their functionality (eg, broken toilet seats, Figure S3-1; broken toilet rolls, Figure S3-2; and destroyed liquid soap dispensers, Figure S3-3).

The majority of toilets had no health related-signage (72.8%). Some form of COVID-19 related health messaging was the most common type (19.5%), followed by the COVID-19 QR code used to facilitate contact tracing via an official smartphone app (12.3%), generic hand washing signage (1.8%) and then non-smoking signage (1.3%), (see Supplementary File 2 for examples of these posters). COVID-19 health messaging signage was more common in the convenience sample than in the three urban council sample (RR = 1.81; 95%CI = 1.21 to 2.71; p = 0.0040). There was no signage that promoted toilet lid lowering prior to flushing.

From a qualitative perspective, we noted that several COVID-19 signs in tourist areas were in Chinese language (Figure S2-1) and a few hand washing signs included te reo Māori wording (language of Indigenous New Zealanders) (Figure S2-3). We also noted that some automatic water and soap dispensers took some time to activate and then dispensed too little soap or too little water for a satisfactory hand wash (ie, repeat activation was required).

Survey results for 2020/21

The comparison of the exact same toilet units involved in the previous survey in 2012/13 is shown in Table 2. There was no improvement in the availability of water for hand washing, but soap availability improved significantly from 59% to 86% (risk ratio = 1.47; 95%CI: 1.25 to 1.72).

Main findings

This survey found a deficient provision of soap (14.8% of toilet units with none), and to a lesser extent water (2.5% with none). These findings suggest that while there has been some statistically significant improvement in soap provision (but not water provision) in the eight year period since the previous survey, the attempts by some local government agencies to increase soap provision at the time of the COVID-19 pandemic¹⁰ needs to be further augmented.

The higher water (5.2% vs 0.4%) and soap provision (21.5% vs 9.6%) in the comprehensive three urban council sample vs the convenience sample respectively, is likely to reflect the more modern facilities in the former and/or a higher quality maintenance schedule. As New Zealand is highly urbanised, we suspect that the true nationwide results would be closer to those of the three council survey than the convenience sample ones.

The relatively low level of health related signage was a problematic finding, especially the COVID-19 related health messaging (at 19.5%), the COVID-19 QR code (12.3%), and generic hand washing signage (1.8%). These deficits are likely to represent both long-term lack of signage relating to hand washing, as well as an inadequate response to the COVID-19 threat.

Also of note from a hygiene perspective was the limited extent of “no touch” activation of devices (ie, for hand washing water at 28.0% and for toilet bowl flushing at 18.5%). These should ideally be expanded with the potential long-term goal being to have maximally “no touch” facilities (including automated: door opening and locking via hand waving in front of a sensor, water dispensing, soap dispensing, toilet flushing and hand drier activation). Automation could extend to the toilet bowl lid being closed before flushing.

Ensuring that all toilet bowls have lids (missing for 32.8% in this survey – often by design) is also desirable, along with messaging to close the lids prior to flushing (see Introduction for the rationale for lid closing).

Study strengths and limitations

This is the largest such survey to date in New Zealand that we know of, and it was able to compare a sub-sample of the same facilities after an eight year period. It was also conducted at a time where there was greatly heightened need for hygiene due to the COVID-19 pandemic.

Nevertheless, the study was limited by only being comprehensive in three contiguous council areas, with the rest being convenience sampling. This was owing to this being an unfunded study with no budget for travel. The convenience sampling is likely to have involved surveying facilities that were more accessible to the researchers by being on main roads. These may be relatively newer and have a better maintenance schedule than facilities in small rural towns. This may have resulted in some under-estimation from the convenience sample of the extent of the problems outside urban areas (eg, with water and soap availability).

Potential research and policy implications

A fundamental research issue is to better quantify the risks of infectious disease transmission associated with use of public toilets (eg, from any aerosolisation of faeces and from touching contaminated surfaces). This is not only relevant to SARS-CoV-2, but also other infectious diseases (eg, norovirus infection and seasonal influenza).

Nevertheless, surveys such as this could be improved upon by making them fully random at the national level and collecting additional data on the facilities to compare them with local standards (eg, as per New Zealand ones,²² albeit from 1999) or more state-of-the-art Japanese designs.²³ Research on ways to minimise the vandalism of public toilets is also needed, as this has been a reported barrier to soap dispensers in New Zealand in the past.¹⁰ Art work inside toilets and on exterior walls is used in some New Zealand toilets (Figure S1-1) and might be worth expanding if it is found to be effective against vandalism.

Policy goals for local government could be to ensure all facilities have water and soap, and to move towards designs that are maximally “no touch”. Built-in redundancy (eg, two separate soap dispensers) may be desirable to minimise the risk of running out of soap, and cakes of soap should probably be avoided as these are more vulnerable to theft. If fully automated taps are not installed, then tap levers or floor pedals for activating water flow could be an alternative. Levers allow users to use the back of their hands and levers can also be more suited than conventional taps for people with disabilities such as arthritis.

Central government could set and enforce minimum standards for council-owned public toilet facilities and the extent of their provision. It could also boost funding support for these purposes, at least partly from border charges collected from international tourists (a funding system already in place in New Zealand). The extent of the funding needed should be seen in the context of the huge costs of COVID-19 and future pandemics, and from other respiratory viruses such as seasonal influenza and from norovirus infection.