Annual outcome.

The annual distribution of the outcome categories is pictured in Fig 5. Most shelter cats were rehomed, with a total average RR of 67.8% (SD = 2.4). The second most common outcome for cats was being returned to their owners (RTO): an annual average of 11.6% (SD = 2.0). Some 4.8% (SD = 1.6) of the cats left the shelter alive for other reasons. For 8.8% (SD = 1.3) of the feline population the non-live outcome was euthanasia or natural death (= mortality rate (MR)). The population Still In the Shelter (SIS-category) consisted of 7.1% (SD = 1.7) of all shelter cats counted on the 31^st of December each year. During the pandemic year 2020 the SIS dropped to a minimum of less than 4% while spiking the next year in the pandemic, in 2021, to more than 10%.

Per individual shelter the annual outcomes differed. Six of the seven shelters showed a decreasing rehoming rate (RR) during the 16 years (Fig 6A). In one shelter the most prominent decrease was seen when its RR reduced from an average of 67% (SD = 4.6) during the years 2006–2013 to 58% (SD = 4.5) during 2014–2021. Only the smallest shelter (dark blue line) showed an increase in the RR. The results from the statistical model (Fig 6B and S3 Table) showed an unstable trend from 2013 onwards and no systematic change could be observed, while in 2010 the RR was 5% higher than in 2013.

The second outcome for cats in shelters is being returned to their owners (RTO). To evaluate the efficacy of this process in the participating shelters, the RTO is presented as a rate of the annual intake numbers of SC per shelter (Fig 7A). Six shelters showed an increase in the RTO over time. The smallest shelter showed a decreasing RTO. One shelter with an average RTO of 26% (SD = 3.1) during the years 2006–2013 increased its RTO to 45% (SD = 4.6) in 2014–2021. The estimated difference in the RTO over time of all participating shelters is presented in Fig 7B (and S3 Table). During the years 2006–2013, the RTO did not change significantly as all confidence intervals covered the zero value. From 2013 onwards, all estimates are positive but still covering zero. Therefore, the chances of a shelter cat being successfully reunited with its owners increased slightly in all shelters to between 19 and 23% (means of seven shelters 2014–2021) but remained quite variable among shelters over the course of time.

The third outcome of live release (average 5.1%, SD = 1.8) represents cats leaving the shelter alive for other reasons. These reasons can be very diverse such as transfer of cats to other rescue centres or in a few individual cases community/feral cats who were neutered and returned to the field.

Euthanasia rate (ER).

The average ER of the total population of shelter cats was 7.2% (SD = 1.3, range: 9.5% in 2008 to 5.5% in 2021). The shelter euthanasia numbers per 1000 human residents (Fig 8A) decreased in those 16 years. In 2006 between 0.08 and 0.45 cats per 1000 residents were euthanised, which decreased in 2019 to 0.03–0.31 cats/1000 residents and dropped to 0.03–0.16 cats per 1000 residents in 2021. The results from the statistical model (see Fig 8B and S3 Table) showed rather stable ERs till 2015 and the reduction was most prominent in the last 3 years, to around 50% of the ER level in 2013.

Death rate (DR).

The DR is the number of deceased cats (by euthanasia or natural death) out of the total feline outcome (but excluding the SIS-category). When calculated over the complete 16-year period, 9.5% (SD = 4.2) of all cats recorded during that time did not leave the shelter alive. The DR proved to be quite variable in the course of time within and between shelters (S3 Fig). After an initial increase the average DR stabilised from around 2009 to 2016 and showed a slight decrease from then on (S4 Fig).

Length of stay (LOS).

The mean LOS for all cats in the seven shelters during the total 16 years was 26 days. S5 Fig pictures the annual LOS of the seven shelters in this study. The smallest shelter in this study has an irregular LOS (S5 Fig–dark blue line) and is therefore excluded from the LOS analysis. The source of shelter cats has influence on the LOS. Small kittens entering/being born in a shelter have the longest LOS: the median was 64 days (mean = 67.6 days, SD = 20.3). The median stay of all SC (including the RTO) was 25 days (mean: 26.0 days, SD = 10.6), while OSCs had a median of 27 days (mean = 28.6 days, SD = 15.1). SCs that were successfully reunited with their owners had a median shelter stay of 3 days (mean = 3.0 days, SD = 1.1).

The mean LOS of its total feline population varied considerably between and within the six shelters: the mean annual LOS of one shelter ranged between 16.0 and 25.0 days in the study period while another showed the highest annual mean LOS of between 23.0 to 39.0 days. Four shelters showed a decreasing trend while two shelters had an increase in the LOS (Fig 9A). Fig 9B (S4 Table) showed an initial increase from 2006 to 2011. From 2013 on the LOS seemed to reduce by 5 to 10%, but this was not statistically significant (except for 2018).

The LOS of all SC (Fig 10A, including SC returned to their owners (RTO)) is less variable among shelters compared with the LOS of all cats shown in Fig 9A. Also, less variability is observed within shelters, although one shelter showed a prominent change from 20–25 days to 5–10 days in later years. On average the years before 2013 showed rather stable mean LOS levels (Fig 10B, S4 Table), while from 2014 onwards the LOS decreased by even 7 days in 2018 and a slight increase during the years of the pandemic.

Data for shelter A (Fig 10A: darker blue line) and shelter B (orange line) show how an increase in the RTO affected the LOS of the SC. The total feline population of shelter A during these 16 years consisted of 70.4% SC. During this study the LOS of all SC (including the RTO-cats) decreased by 36% from 33 days in 2010 to 21 days in 2020 and 2021, while their RTO increased from 19% in 2006 to 25% in 2021. Shelter B, with a comparable feline population of 69.8% stray cats, had a gradual increase of the RTO from around 20% in 2008 to around 45% in 2019–2020. During the same time the LOS of SC (including the RTO-cats) was reduced by 80% from 25 days to 5 days in 2019 and 2020.

Risk-based live release rate (RLRR).

The RLRR reflects the probability (= chance) of cats being released alive from the shelter during a certain time interval, as the denominator of the RLRR contains all cats with and without an outcome in the shelter (= SIS-category).

In 2006 the RLRR varied between 78 and 87% and increased to 85–93% in 2021 (Fig 11A). One shelter (yellow) had a considerably lower RLRR compared with the others during the first years, but in later years this increased to the same level as the rest. However, more variability in RLRR is seen within shelters than between shelters. From 2006 to 2012 the mean RLRR was comparable with that of 2013 (Fig 11B) and between 2014 and 2021 the mean RLRR increased by 5%.

Still in shelter (SIS).

Being no part of the outcome, the SIS-category is a specific metric. During the 16 years the SIS remained quite stable, with an overall percentage of 7.1% (SD = 1.7). During the pandemic the SIS dropped to 3.8% in 2020 and increased to 10.1% in 2021. When excluding the pandemic years 2020–2021, the average SIS was still 7.1% (SD = 1.4).

Rehoming rate (RR) and return to owner (RTO).

The two major live outcomes were rehoming and RTO. Most cats were rehomed, a process dependent on adequate presentation of the animals to the public [36, 37] and other factors including animal health. Six out of seven shelters showed a gradually decreasing RR. The reasons for this decrease could be caused by a combination of factors, such as the increased RTO. RTO is the second live outcome and can be calculated in different ways: as part of the total intake [29] or related to the number of SC. In this study the RTO is a function of the total stray intake, as recommended by Scarlett [14]. As fewer SC and OSC were taken in by Dutch shelters, while RTO was increasing fewer shelter cats were available for rehoming. An increase in the RTO has a direct dampening effect on the RR. One shelter showed this effect most prominently: with an increasing RTO of 20% of the SC in 2008 to 53% in 2018, its RR decreased from almost 75% in 2009 to 50% in 2018. This shelter explained its RTO increase by an effective use of social media for missing and found pets and being available 7 days a week to allow owners to pick up their stray pets. As Hawes et al. [26] suggested earlier “whether decreased intake of stray animals by these facilities is associated with increases in a community’s use of non-sheltering methods (e.g., social media and mobile applications) to return lost companion animals to their homes”, the data in this study seem to support this assumption.

Following the advice of Scarlett [14], in this study the RTO was calculated as part of the SC (and not as part of the total number of cats taken in) because only stray cats can be returned to their owners. In this way, a shelter can closely monitor its success in returning stray cats to their rightful owner. Shelters in this study showed remarkable high RTO rates when compared internationally. However, this could be caused by differences in definitions. The RTO as a proportion of the total intake (with a larger denominator) is smaller in contrast to the RTO as part of the total stray intake. These differences in the calculation of the RTO make scientific comparison difficult and do not stimulate improvements made by individual shelters. Following Scarlett’s lead [14], we therefore recommend calculating the RTO as part of the SC rather than the total feline intake.

Rates of mortality (MR), euthanasia (ER) and death (DR).

In several publications death rates were determined solely from the group of animals with an outcome (since passing away is an outcome), without acknowledging the risk of dying for every single shelter animal regardless of whether there is an outcome for that animal. The correct way to analyse the mortality in a shelter therefor is by using the MR and ER as these rates not only include cats with an outcome, but also the SIS-part of the population [14, 38]. Earlier European research [16, 17, 39] frequently used the DR instead. To be able to compare the Dutch shelter metrics with this research, the DR was therefore included in this study. Dutch shelters showed a DR of 9.5% (SD = 4.2) of the total feline outcomes during these 16 years. This is significantly lower compared with the DR of 10 Belgium animal shelters in the Brussels area [16]. In 2018 Brussels shelters had DRs of 29%, and in 2019 27%. In our study, only in 2009 did one Dutch shelter report a DR comparable with the Belgian shelters, at 22%. In 2020 the Belgian DR decreased considerable to 11%. In addition, three Czech cat shelters reported DRs of 33% between 2011 and 2015 [17].

The ER of Belgian shelters was also higher compared with the overall average Dutch ER of 7.2% (of the total population): 19 and 15% of Belgian shelter cats were euthanised in 2018 and 2019. The pandemic probably influenced these numbers in Belgium: in 2020 the ER dropped to 4%. The shelter metrics of German shelters are comparable with the Dutch: in 2011 Arhant et al. [39] reported ERs of between 5 and 10% of cats entering German shelters. Following the US methodology, our study standardised euthanasia cases per 1000 human residents. In 2006 the total euthanasia rate of Dutch shelter cats was below 0.60 cats/1000 residents and in 2021 it dropped below 0.20 cats/1000 residents. Although Hawes et al. [25] reports an almost 80% decrease in ERs at shelters in Colorado (from 5.8 cats/1000 residents in 2008 to 1.2 cats/1000 residents in 2015), the Dutch ER is considerably lower. Cultural differences (Europeans prefer an indoor/outdoor life for their felines, while US pet owners prefer their cats to stay indoors [40]), differences in legislation (protection of shelter animals by US legislation is organised at state level [41] and might differ locally [42]), as well as differences in the size of the cat populations (US: 290 cats/1000 residents [40]; the Netherlands: 175 cats/1000 residents) might influence the ER, but more research is needed to explain these differences.

All participating shelters were transparent in providing mortality data on natural deaths and euthanasia of shelter cats. The public perception of good and compassionate care for shelter animals does not always involve euthanasia [43]. When public opinion concerns prompt shelters to under-report this data, it will complicate the analysis and management of potential risk factors for euthanasia [17, 44]. Although it can damage a shelter’s reputation, correct reporting of this data is essential.

In individual shelters mortality (MR) and euthanasia rates (ER) can shift with time. When a shelter decreases its ER (an aim for many shelters) but shows no decrease in its MR, this can be caused by a shift from less euthanasia to more natural deaths. When a delay in euthanasia results in more natural deaths, this could be an indication of welfare problems [14], as euthanasia can shorten the suffering of seriously ill cats. However, in this study we only compared trends in ER without the relation with the MR and further conclusions cannot be made.

Length of stay (LOS) and return to owner (RTO).

This study showed the relation between the LOS and the RTO cats. RTOs only spent an average of 3.0 days in the shelter, while rehoming the rest of the SC is delayed by the mandatory 14 days of legal holding [3]. Although increasing RTO requires shelter time investment (checking registration of lost pets is labour-intensive), these efforts can reduce the LOS for the total population of SC. Lowering the LOS decreases the risk of cats developing upper respiratory tract disease [11, 45]. The LOS can therefore be a(n) (indirect) tool to monitor feline welfare. When the RTO increases substantially, in shelter metrics it can be useful to indicate the RTO population separately from the total SC population to better monitor the LOS of the rest of the non-RTO-SC.

This steady increase of the RTO, however, might also give a subtle warning. Do these cats, taken in as strays, really need 3 days of shelter facilities before being returned to their owners? Or do we need a new approach to avoid an increased feline circulation through our shelters, with all the risks for the cats involved? Hurley recommends a triage system for shelters at the intake of a stray animal: “By transporting found animals to a shelter without first making efforts to locate the owner, well-intended finders often unwittingly reduce the likelihood of the pet ever being reunited with its family” [46]. Although the RTO is rising, the so-called ’stray but well-cared for outdoor cat’ risks losing its caretakers when the animal is taken to a shelter, while it could have managed well in its own environment [47].

Risk-based live release rate (RLRR) and its influencers.

The probability of a Dutch shelter cat being released alive has substantially increased during the past years, to a nine out of ten chance. As the live release in the RLRR (= nominator) consists mainly of two components–RTO and rehoming–it is not affected by a shift between those two outcomes. RTO and rehoming merely act as communicating vessels within the RLRR: the feline outcomes may shift from one to another (more RTO, less rehoming, etc.), but cats still leave the shelter alive. Two other developments, however, substantially impacted the RLRR increase: a decrease in feline intake and a decrease in the ER. The decrease in feline intake could stimulate the rehoming of relatively more elderly and less robust animals (e.g., diabetic or hyperthyroid cats, cats infected with feline retroviruses (FeLV/FIV). When these cats are rehomed instead of euthanised, the RLRR increases. This trend requires improvements in veterinary guidance and animal care and therefore has implications for the level of shelter medicine.

The denominator of the RLRR includes all cats with and without an outcome (the SIS-category). These two parameters can reinforce each other. When cats with a live outcome are leaving the shelter fast (= decrease of the LOS), the SIS decreases, resulting in an increased RLRR. With a lowering of the LOS, the risk of infections decreases as well (for upper respiratory tract disease (URTD) [45]), resulting in a decreasing morbidity and mortality, also resulting in a higher RLRR.

From approximately 120 shelters in Netherlands, a convenience sample of seven shelters was taken which voluntarily participated in this study. Obtaining real data from individual shelters proved to be somewhat challenging. The criterion for participating in this study was the availability of digital shelter data over a time period between 9 and 16 years. Several shelters were therefore unable to join owing to a lack of digital data or too short a time span of the data. A selection bias of this convenience sample can therefore not be excluded. While most shelter metrics were well defined, an information bias was introduced by a variable where each shelter used its own time interval to define returned shelter cats as "returned adoptions". This variable was therefore not included in the analysis. Unfamiliarity with participating in external research projects and sharing the results publicly may also have played a role in some shelters’ reluctance to participate. If there had been a central Dutch registration of admission and outcome of shelter animals, a scientific analysis would probably have been carried out earlier with data from more shelters.

It is important for future research that more shelters participate in comparable studies to be able to verify the observed trends [48], to study existing metrics and develop new statistics for optimising shelter management. Also, the animal species in these studies should not be limited to cats but should include dogs and other species as well. Furthermore, the impact of pet rehoming services and TNR programs for stray or feral cats on the size of shelter populations should be explored.