What Effect do Kittens Have on the Adoption Rates of Non-Kittens?

For high volume shelters, moving animals out the door as quickly as possible is crucial, and running a successful adoption program is one way to achieve quick movement. There is a plethora of ways to create adoption success: strong adoption promotion, open adoptions procedures, and reasonable adoption fees. But, have you ever thought about what impact the mix of animals in the adoption room has on the overall adoption rates? Specifically, does the presence of kittens help or hinder the adoption rate of older cats?

Dr. Kate Hurley and Dr. Sandra Newbury of the UC Davis Koret Shelter Medicine Program (http://www.sheltermedicine.com/) have lectured and researched this concept. According to their lectures, in adoption rooms across the nation “the more, the merrier” notion is often employed without much thought to animal presentation or organization. We pack as many cats into the adoption room as possible, and leave it up to the animals to sell themselves. Those shelter animals are so darn adorable and do a great job of it, but research and data are actually showing that a little forethought and planning can improve adoption rates. In his book, The Paradox of Choice: Why More is Less, author Barry Schwartz argues that eliminating consumer choices can greatly reduce anxiety for shoppers. So, extrapolating this concept to an adoption room is another way that animal shelters can work to improve their adoption rates.

To investigate one aspect of this concept, I looked at data for a medium open-admissions shelter in a large city. As mentioned above, I was curious to know if the adoption rates of adult cats rise or fall when kittens are also available for adoption. One might argue that having kittens in the adoption room will increase older cat adoption rates because kittens are like ‘eye candy’ and draw-in more would-be adopters. On the other hand, one might argue that kittens decrease the adoption rates of older cats because kittens are inherently more desirable.

A few notes about these data:

1. Kittens are 4 months and younger; non-kittens are 5 months and older.
2. Two months were chosen because kitten intake rates fluctuate seasonally (February = lower kitten intake; August = higher kitten intake) so looking at two different months can help account for this fluctuation.
3. If no kittens were available for adoption on any day in the month, the day was removed from the dataset, thus there are less data points than days of the month.

That being said, let’s look at the data:

Analysis:

We can ascertain from the negative slope of the trend line for both months that having kittens available for adoption actually detracts from adoption rates of non-kittens. But, the negative slope improves in August—a time when kitten intake is at its highest:

• In February, for every 4 additional kittens available, this shelter looses about 3 non-kitten adoptions
• In August, for every 3 additional kittens available, this shelter loses about 1 non-kitten adoption

So, does this mean that shelters should not make kittens available for adoption? Certainly not! But, analyzing data in this way can be the first step towards understanding adoption patterns and choices. Knowing that kittens will take away from adoptions of other cats will help shelter managers space kitten arrival in the adoption room or increase their use of foster-to-adoption programs (an emerging trend in animal sheltering to “foster” underage kittens to families with the eventual hope/mutual understanding that the family will adopt the kitten when it is of age) for young and underage kittens.

Moving forward, shelters can analyze their data in this way for various types of scenarios including:

• Adopter preferences of males vs females
• Adoption rates of different colors (black vs. non-black) or coat patterns (solid vs. tabby)
• Adoption rates for various kennels or rooms housing available animals (do animals housed in the kennels near the front of the room get adopted at different rates than those housed near the back?)

And, to further analyze this topic, I would look at each month of the year separately, and then the entire year as an aggregate. In addition, I would play with the age of kittens a little to see if the data change in any way. For example, define a kitten as anything under 6 months old, or twelve months old. And, then further split the age groups: kittens vs. seniors, kittens vs. young adults, kittens vs.adults, etc...

Let me know if you have any questions about this particular analysis, I’d love to hear how your shelter’s data stacks up!

The 7 Habits of Highly Effective Shelters (Statistical Version)

Ah yes, another take on the infamous “7 Habits…”!  Below are some thoughts on why statistics are just as important as good husbandry techniques for adequately managing a shelter population.  Statistics are more than just an afterthought, or “I’ll get to them as some point” kind of mentality, and need to be added to the forefront of shelter operations.  Capturing good data and then using that data to plan and adjust to changes in your population will help your shelter go the extra mile in every aspect of shelter management possible.  Admittedly, the following is geared towards brick and mortar animal sheltering facilities, but I think any type of rescue organization can relate to these ideas.  Read on to explore the 7 Habits of Highly Successful Shelters (the Statistics Version):

1.  Garbage In-Garbage Out.  The single most important factor in using statistics to your advantage is to gather good data in the first place.  Don’t wait until later to record an intake, medical exam, outcome, behavior test, etc.  Do it now, and do it correctly.  You have no hopes of your software producing good reports if the data going in are incorrect or missing.  Shelters are so dependent on measuring time and making sure that animals move through quickly and efficiently, therefore it is vital to record events and observations AS THEY HAPPEN.  You are dead in the water if you are not recording data as events occur.
2. You have to do a regular head count.  Depending on the capacity of your shelter, this might mean every day.  This can be accomplished when shelter managers/veterinarians are doing rounds, or when staff are going through the rooms each morning for first feedings, or this can be one person’s job to do each and every time.  I have even heard of shelters using bar codes and scanners to take inventory counts!  Whatever works best, just make sure it happens.  There is no way you can produce good statistics, if you are not 100% certain of the animals in your population and where they reside at every point in time within your facility.  As an added benefit, this will also help to reduce ‘lost” or “missing” animals. 
3.  Know your carrying capacity and its limits.  By this I don’t mean know how many cages you have; carrying capacity is more than just number of cages.  Carrying capacity means calculating your intake “comfort zone” based on your usual “outcome activity”, rather than the other way around.  It means adjusting to unusual intake patterns such as kitten season or puppy mill raids to account for the needier (ie, resource draining) events/populations.  It means having an action plan in place for when you are reaching that carrying capacity—watching statistics on a weekly or monthly basis will allow you to be proactive and notice even subtle trends/changes in your data.  Managed/limited admissions facilities will find this task slightly easier to accomplish than open admissions shelters, but that just means those shelters will have to be a little more creative and flex a little more mental muscle.
4. Set benchmarks. Once you are consistently gathering quality data, go ahead and set benchmarks (ie, goals) for the statistics that you most closely watch.  This will give you even further insight into the health and well-being of your shelter.  Are you consistently meeting your benchmarks?  If you are not, what action items will be put in place to adjust for the discrepancies?  Increasing the benchmarks themselves does not count!  Benchmarks will also provide direction for staff and hopefully increase productivity if they have in mind what goal they are working to achieve.
5. Share your stats with staff.  This is important for a couple reasons.  First, it’s always good practice to have many eyes looking at the data to notice any discrepancies, errors, and to make observations/interpretations.  But in addition, this is a great way to give all shelter staff an appreciation for their work beyond direct contact with the animals—in other words, this will give them a glimpse at more of the “big picture”.
6. Talk with your shelter management software company, and often—you do have a specific animal shelter management software company, right? (We need to talk, if you don’t, this is an integral part of keeping good data)! Tell them what you need and why.  Maintaining an open line of communication with this group will ensure that your data management becomes a fluid, efficiency creator for your daily operations.  Furthermore, it has been my experience that many of the developers of these types of software are computer people and not “animal people” or “helter people”.  So, from their perspective, the software is performing just perfectly, but from your perspective—someone in the trenches daily—you might think the software needs a little tweaking.  Go ahead and speak up!
7. Collect basics statistics and report on a regular schedule (weekly or monthly is preferred).  Every shelter regardless of size should at least be collecting the following, and be able to report it by variables including animal type, animal age category, etc: a. Intake numbers, b. Outcome numbers, and c. Length of Stay/Animal Care Days.  Basically, you need to know what is coming in, what is leaving, and how long they are staying.  I would guess that about 80% of all meaningful shelter statistical analysis would in some way include these three stats in the respective calculations, so make sure you start with these.
   
   

Tracking Length of Stay Statistics

*How will this kitten group affect the overall shelter's LOS? 

While we have discussed length of stay (LOS) in a few different contexts already, we have yet to talk about the myriad of ways it can be calculated and what information each of those calculations lends to managing an animal populations.   Similar to calculations for human patients in a hospital, LOS in its most simplistic definition is the intake date minus the outcome date, and again related to hospital computing, if a group finds itself lucky enough to have a relinquished animal come in and leave within the same day, I always advise groups to count that animal’s LOS as one day.  Individual animals have personal lengths of stay and groups of animals contribute to average lengths of stay.  It is within this latter aggregate group that the power of the LOS statistic shines through. 

The first step in gathering meaningful information from aggregate LOS is to define cohorts of animals you are interested in analyzing.  Presented below are some suggestions for analysis and descriptions of what benefits the data offer.  All the calculation proposals below can also be subdivided into animal groups such as all dogs, all cats, all animals, just puppies, just kittens, etc.

1. Intake Cohorts:  Calculating an average length of stay for animals grouped together by the month (or week, or day, etc) they came in to the shelter.  Tracking LOS in this manner will give you insight to any seasonal variations in your shelter’s operations and how your population management handled things.  Does the January Intake Cohort move through quickly compared to the July Intake Cohort?   Using historical LOS data from last year’s Intake Cohorts, shelter managers can make decisions about enlisting additional resources such as foster homes, staffing, or volunteers to help alleviate anticipated increases in LOS during particular time periods.  One limitation to using this calculation is that you might not get stable data until many months  (or weeks, or days, etc) after the intake month under consideration has passed.  If your group routinely moves animals through within 15-20 days, then this will not be such a problem, but if your shelter does not, then it may take a while for all members of this cohort to have an outcome, and have their data contribute to the calculation.   For example, if the shelter manager was calculating the February Intake Cohort’s LOS, he or she might have to wait until April to do so as those animals that came into the shelter in late February will need time to pass through the system.
2. Outcome Cohorts: Calculating an average length of stay for animals grouped together by the month (or week, or day, etc) they left the shelter.  This can show insight into the health of a shelter’s adoption and/or transfer program.  And, opposite of the Intake Cohort limitation mentioned above, Outcome Cohorts can be calculated immediately for their specified time period.  For example, if a shelter manager was calculating the February Outcome Cohort’s LOS, he or she can do so on March 1.
3. Current Cohort: Calculating the average length of stay for all animals at your facility at this moment in time, regardless of intake date.  This is probably as close to real-time shelter data as a shelter manager can get!  Again, this particular calculation can give insight into the shape of your adoption/transfer program, but can also give a shelter manager clues about the overall well-being and health of the population.  Animals that are ill or need extended care will increase this calculation, whereas a healthy population will move through the system quickly and therefore have a shorter LOS.
4.  Stage/Location Cohorts: It is also useful to calculate a LOS statistic for animals in specific stages or locations within a shelter system.  Much like the calculation for Current Cohorts, monitoring the data from a Stage/Location Cohort will ensure that bottlenecks for whatever reason are noticed and addressed before they affect the greater population.  Examples of possible stages and locations to monitor include:
1. Intake/Processing
2. Medical
3. Holding
4. Adoption Rooms
5. Foster Stays
6. Rehabilitation Rooms
5. Intake Type Cohorts: Calculating LOS for intake types such as owner guardian surrenders, strays, transfer animals, or emergency response intakes, can give a shelter manager understanding of the burden that these subcategories of animals place on the global shelter system.  For example, if a shelter takes in a large transfer group from another shelter or even a hoarding situation, it will be useful to track both the transfer group’s LOS, as well as all other animal’s LOS.  If we see a rise in the LOS for all other animals (compared to data for this group from pervious months or time period), that is an indication that perhaps the transfer group was too burdensome or overwhelming for this shelter’s operations as it affected the data for these other animals in the shelter’s care at that time.  From this, shelter managers can create operation protocols for dealing with such burdens so that the rest of the population is not affected.
6. Outcome Type Cohorts: Calculating LOS for outcome types such as adoption, transfer out, euthanasia, etc.  This will inform shelter staff as to the efficiency of daily operations, as well as the health and well-being of their population.  For example, if a shelter euthanizes animals, it is imperative to watch this LOS so that efficiency is managed.  Although euthanasia is not always a predictable outcome and shelters would like to give all animals as much of a chance as possible, if Outcome Euthanasia LOS increases, it might be indicative of unnecessarily delaying inevitable outcomes.  

The second step in monitoring LOS statistics is to set goals or benchmarks for each of the categories of data a shelter is tracking.  I would recommend monitoring data for a year before determining baseline statistics.  Although one year seems long, you must account for seasonal fluctuations and therefore need the entire year to see a complete picture of your shelter.  Once baseline is established, go ahead and set goals for each category.  This will help focus operations and will certainly drive decisions made around the shelter.  And again, based on any seasonal fluctuations recorded in your baseline data, it would be good practice to set fluctuating benchmarks to accommodate the ups and downs.  All of us shelter workers know all too well what the warmer months (i.e., “kitten season”) can do to a shelter’s numbers, so it is acceptable to adjust your expectations at this time based on what your baseline is telling you.

What types of LOS does your shelter track?  Has knowing your LOS been beneficial to your organization and staff?  Does your shelter set goals or benchmarks?  What happens if you do not hit your benchmark?

*Photo courtesy of Chris Tanaka

Calculating Staffing Requirements Based on Animal Population

I don’t think anyone would argue if I made the blanket statement that all shelters are on a limited budget and must carefully apportion their resources in order to manage their animal population best.  In that spirit, one of the most important resource allocations a shelter makes is deciding how many staff are required to clean and feed the animals in its population.  Too many staff can drain the shelter’s budget, and too little staff can compromise the health and well being of the animals in its care.  So, how does a shelter manager decide how many staff to assign?

To answer this question, let’s start with the animal care time requirements suggested by HSUS and NACA and also endorsed by the Association of Shelter Veterinarians in their 2011 Guidelines for Standards of Care in Animal Shelters document.  These groups estimate a minimum of 15 minutes is required for adequate animal cleaning and feeding.  This is a good starting point for calculations, but if you feel 15 minutes is too long or too short, you can adjust the calculations accordingly to hit your goal.  Next, we need to know how many animals are in the shelter.  We can use our historical data to make staffing projections, or we can use the daily shelter count to make real-time decisions.  Let’s use some faux data from Shelter XYZ:

Table #1: Shelter XYZ Daily Population, January

January	Total Cats	Total Dogs	Total Population
1	100	75	175
2	111	76	187
3	99	79	178
4	98	80	178
5	99	81	180
6	100	76	176
7	104	74	178
8	109	70	179
9	121	69	190
10	123	68	191
11	120	66	186
12	121	67	188
13	125	69	194
14	121	61	182
15	119	63	182
16	118	65	183
17	115	69	184
18	113	64	177
19	101	70	171
20	99	71	170
21	99	60	159
22	100	61	161
23	111	62	173
24	111	63	174
25	112	64	176
26	108	66	174
27	106	60	166
28	103	70	173
29	102	71	173
30	101	77	178
31	99	70	169
Average:	108.6	68.9	177.5

And, finally, the third variable of total time allotted for staff to finish the task is needed.  For our calculations below we will use four hours, but based on each individual shelter’s operating procedures, this may differ.  

To calculate for real-time staffing decisions (using January 21^st as an example day):

• We had 159 total animals in the shelter that day: 159 x 15min/animal = 2,385min
• 2,385min / 60min/hr = 39.75 care hours
• Staff are expected to finish cleaning and feeding in 4 hours: 39.75 care hours / 4hr = 9.9 (10) staff required to complete the task

To calculate for projection staffing (using January animal averages):

• On average, we had 177.5 animals in the month of January at Shelter XYZ: 177.5 x 15min/animal = 2,662.5min
• 2,662.5min / 60min/hr = 44.3 care hours
• Staff are expected to finish cleaning and feeding in 4 hours: 44.3 care hours / 4hr = 11.0 staff required to complete the task

What if you are curious to know how long it is currently taking your staff to clean and feed the animals?  One option is to time your staff.  But, I would not recommend this as anyone doing a task knowing they are being times/watched will at the least be suspicious, and probably perform differently as a result—in other words, your data will be significantly skewed.  The other option is to work the above math backwards to find a time per animal.   For example, if I have five staff cleaning and feeding for three hours, and 50 animals they are tending to, then they are spending 18 minutes per animal:

·      5 staff x 3 hrs x 60min/hr = 900min

·      900min / 50 animals = 18min/animal

Although I am not a fan of timing staff, now knowing the theoretical time per animal, a shelter manager can pull out their stopwatch to see how the data compare. 

Looking at the population numbers from the past year will allow shelter managers to plan appropriately for the upcoming year, and help save money and stress when high population season hits.  For example, if we know from last year’s data that June-October produces our highest populations, then the shelter manager can prepare for the staffing fluctuations needed to accommodate the rise in animals in February or March, rather than be taken by surprise in June and July and have husbandry and animal health compromised as a result.  When hiring seasonal help, I have found two techniques to be of great service:

1. Begin the hiring process 2-3 months before you actually need the seasonal help.  This will allow you to train new staff during a period of lower population when stresses are less and you can dedicate more time and attention to training.  When the high season begins, they are all in place and ready to jump in.
2. Consider rehiring the same seasonal help year to year.  This works great if you have a relationship with college students looking for summer work or even stay at home moms or retirees who would like a little work from time to time.  And again, having the same seasonal help will minimize the learning curve.

Photo from Health Technology Professional Products (htproducts.net)

Once you do the calculations above, and set your goal for animal contact time, what should a shelter do if it can only employ so many paid staff?  Using the data from January for Shelter XYZ above, we need 11 people to accomplish the task in four hours.  Shelter XYZ only has budget for eight staff.  My suggestion is to get trained volunteers in to work along staff and fill in the staffing gap to hit the target.  

Tracking Animal Transfer Placements

One of the many benefits that has emerged from the growing cooperation of humane societies and animal welfare organizations in the past decade is the concept of transferring animals from one organization to another.  Many open admissions and municipal facilities depend on transfer programs to increase live release rates, and most smaller rescue and limited admissions shelters only pull animals from other facilities.  For everyone involved—large and small organizations, municipal shelters, rescue organizations, and of course, the animals—it is a win-win situation.  Additionally, a transfer program also addresses pet distribution inequities.  In Chicago, we are lucky to have a network of hundreds of breed and specialty rescue organizations all working to save more lives; if an adopter is unable find what they are looking for at one shelter, they can readily open the latest addition of Chicagoland Tails to browse the vast rescue list and check out other organizations to find their new family member.  Some shelters have relationships with organizations in other states and even other countries.  Click here to read about a creative transfer program to get island pets to the mainland.   

Photo from Boggs Mountain Humane Shelter: boggshumaneshelter.com

Now that the transfer concept is established and working well in most communities, it is time to add sophistication to the data analysis for transfer programs.  Most shelters will calculate transfers as a variable in their live release data—which is good, but at this point, not enough.  How does a shelter know what actually happens to the animal once it is transferred?  From personal experience and anecdotal data, I know that many shelters keep in touch with their transfer partners and share photos and updates on the animal, but so far, not many have been recording actual data on placement rates, lengths of stay, or any other information.  I know, I know, one more thing to add to the shelter’s ever growing list of important things to do.  Here are a few suggestions:

1. This could be a great volunteer job—from your shelter software, print out the current list of transfers to other organizations, give the lucky volunteer a telephone, and let the calling begin.   
2. Be proactive about the process—add a clause in your transfer contract that puts the responsibility on the end of the transfer-receiving group to notify you when an animal is dispositioned. 
3. Of course, use your already existing shelter software to manage the data—if there is not a field or page in the software to add details to outcomed animals, talk to your software administrator to work on a solution to make this easy for you!  (You can even point them to this post to convince them of its necessity!)

And, here is a list of possible data to track:

1. Length of stay at transfer receiving organization—like length of stay within your own organization, this will give you a good overview of that organization’s overall health and adequacy at placing animals. 
2.  Final outcome percentages—are all animals adopted?  Some may be transferred yet again, some may be euthanized, some may become permanent residents of the receiving organization (sanctuary-type situations). 
3. Disease rates—did said animal get sick?  On what date of care did it break with symptoms?  What are the symptoms/diagnosis?  This will be helpful in determining your own shelter’s disease rates—if the animal broke with symptoms within the first few days of transfer, you can make the assumption it contracted disease while in your care and subsequently, you can add the data to your own disease calculations.  If it broke with symptoms well after transfer, then the disease transmission likely occurred while in the receiving transfer organization’s care.
4. Behavior Notes—was there unexpected behavior training/modification necessary?  If so, of what nature?  Did the behavior issues prohibit or increase the time to adoption? 

Following these numbers will give your organization a better understanding of its transfer program, and also strengthen the relationship with your transfer partners.  After all, in its broadest scope, the goal of a transfer program is to initiate more and more appropriate animal placements, so knowing your data (ie, your transfer partner relationships) more intimately will assist in successfully arriving at this goal.  You will have a clearer understanding of which animals to place with which groups, which groups to work with on a regular basis and those to work with on an episodic time frame, and you can also justify pulling away from certain groups if the data does not support the relationship any longer.  Further, if you notice lengths of stay increasing for one group or all transfer partners, you may look more closely at your own data to find out what is causing the swell.  Perhaps, your husbandry practices are slipping, and the animals transferred out are requiring unexpected or additional medical interventions thus increasing lengths of stay at the receiving transfer organization.  Or, maybe the issue is not with your group, but with the receiving transfer organization, and you may then consider advertising those animals on your own shelter’s website to boost exposure to potential adopters.  (Word of caution: *so as to not confuse or frustrate adopters, clearly label transferred animals as not being in residence at your shelter).  Whatever you decide to do with your transfer program data, I encourage all shelters to take this next step in enhancing the animal transfer concept and overall adding to the benefits for the animals.  Hats off to The ASPCA, whom in their $100K Shelter Challenge began requiring transfer affidavits in order for any transferred animals to count as a successful live release during the challenge period!  Click here to read their transfer rules (scroll down to “Challenge Rules” heading, point #9).    

Let’s Roll With It: Calculating Live Release Rates

The live release rate is a very telling metric in measuring a shelter’s progress towards improving dispositions for the animals in its care.  Literally translated, a live release is any animal that leaves your shelter alive.  Options for live release include:

(1)  adoptions

(2)  return to owner (RTO)

(3)  transfers

(4)  TNR—or feral colony placement

Simply calculated, the Live Release Rate (LRR) is the sum of all the live outcomes for a period of time divided by the total intake numbers for that same period of time. 

Let us consider some faux data from Shelter XYZ:

	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sept	Oct	Nov	Dec
Total Adoptions	97	88	99	100	78	84	94	99	102	87	85	105
Total  Transfers	12	13	8	5	11	10	12	9	4	11	5	6
Total RTO	5	19	3	10	2	4	2	15	18	1	11	2
Total Intake	250	249	234	112	276	270	265	245	235	241	223	256

Calculating the LRR for January: (97+12+5)/(250) = 45.6%

LRR for February: (88+13+19)/(249) = 48.1%

LRR for March: (99+8+3)/234 = 47%

The astute reader will wonder what is going to happen with April’s calculation?  Well, let’s see:

LRR for April: (100+5+10)/(112)= 100.2%

What?!?!  How is it possible to have more animals leave the building alive than what we took in that month?  The math does not make sense.  Enter the Rolling Live Release Rate (RLRR).  The RLRR allows the user to adjust for a holding or starting population of animals that carry over from month to month. 

Let’s consider our data from above with additional rows of information:

	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sept	Oct	Nov	Dec
Total Adoptions	97	88	99	100	78	84	94	99	102	87	85	105
Total  Transfers	12	13	8	5	11	10	12	9	4	11	5	6
Total RTO	5	19	3	10	2	4	2	15	18	1	11	2
Total Holding	45	30	42	32	29	28	29	25	46	44	41	39
Total Intake	250	249	234	112	276	270	265	245	235	241	223	256

Now, calculating a RLRR for January: (97+12+5)/(45+250) = 38.6%

RLRR for February: (88+13+19)/(30+249) = 43%

RLRR for March: (99+8+3)/(42+234) = 39.8%

Now, let’s examine the calculation for the odd month of April:

RLRR for April: (110+5+10)/(32+112) = 79.8%

Here is how the numbers compare:

	Jan	Feb	Mar	Apr
Live Release Rate	45.6%	48.1%	47.0%	100.2%
Rolling Live Release Rate	38.6%	43.0%	39.8%	79.8%

Why would I want to produce RLRR when they make my data look worse?  Because you must account for the carry over population for each time point (monthly, annual data, etc).  I have yet to encounter a shelter or rescue that can disposition all of its animals immediately; inevitably there is some carry over population of animals no matter how talented and speedy the shelter is in live release placements.  In a rolling live release the denominator is the sum of all animals that have the potential to be outcomed, so one must incorporate the holding population in the computation.  And, as is the general rule for most statistical markers, it’s the trend that matters, not the absolute numbers themselves.

RLRR can also be calculated for subpopulations within your shelter: cats and dogs, puppies/kittens and adults, male and female, and so on.  In these calculations, you would only sum the total live outcomes for the subgroup under investigation (puppies, kittens, etc) divided by the total potential for outcome in that subgroup.  In order to not underestimate the RLRR, always remember to divide by the subgroup, not total intake—in this example, all dogs and cats, etc).