Factors that may contribute to SA

Through characterization of the various pheno-types of SA, factors, including signalment, that contribute to the risk of developing SA may be identified. However, results of most studies^12,14,,27,28 are based on owners' answers to various questionnaires, with some studies that include a variant of a more objective clinical questionnaire. Also, most studies do not rely on direct observations, so some results may be inaccurate (false negative results more likely than false positive results because owners have not observed abnormal behaviors). If the questions are sufficiently detailed and responses are specific, few to no false positive results are reported.^21,22,29,30 Studies consist of various subjective and objective assessments and clinical observations, such that areas of insufficient knowledge are highlighted and attributional and type I errors regarding causal factors are possible.

Studies were identified using the terms separation anxiety and dogs, separation distress and dogs, separation related distress and dogs, canine separation anxiety, dogs alone, canines alone in Google Scholar and PubMed between June and July 2020. Evaluated possible causal factors of SA were sex; reproductive status; age of acquisition, onset, and referral for behavioral examination; previous ownership; number of people and other pets in the household; owner gender; owner perception of their dog's behavior; source of dog; participation in an obedience class; boredom; exercise, trauma and change in routine; cognitive bias, excessive emotional connection and praise or overindulgence (spoiling); temperament, personality, and breed; genotype; and attachment. Possible causal factors, whether an association exists between these factors and SA, and the strength of the association are summarized (Table 2).

Sex—Although many studies^{15,17,30,31,32,33} show ten dencies of overrepresentation by 1 sex, numerous researchers have failed to identify any risk or reporting difference attributable to sex. Because selection of dogs for enrollment was not random for most of these studies, bias associated with sex may be present; therefore, caution is recommended with interpretation of study findings. Yet, studies by Podberscek et al,⁸ King et al,³⁴ and Flannigan and Dodman⁹ had more male dogs and the study by Palestrini et al²³ had more female dogs, their findings did not indicate sex had an effect on the development, presence, or treatment of SA. Owners of male dogs may seek help earlier because males reportedly have more destructive behaviors,⁹ which may be size or activity related, but data that suggests this are unreliable.

Reproductive status—Most studies^9,31,32,33 reveal no association between reproductive status and SA, but the lack of an association may be because of small sample size. Data are inadequate to fully evaluate the effects of neutering on SA because many dogs in these retrospective studies were already neutered when they showed signs of separation-related distress. An increase in prevalence of SA and other behavior problems among neutered dogs is noted in 1 study¹⁶; however, most dogs were neutered and temporal order (ie, behavior problem preceded neutering or vice versa) was a confounder.

The veterinary experience (perianesthesia, surgery, and recovery) associated with neutering may be stressful and increase the risk of problem behaviors for a dog with cofactors, including genetic cofactors, to develop SA.¹⁶ However, the effects of the veterinary experience on dogs have not been evaluated, such that a well-designed study to identify the effects is necessary.

The presence of progesterone in female dogs has been speculated to have a calming effect; therefore, ovariectomy could predispose a dog to anxiety-related conditions. A prospective, randomized, controlled, blinded study is needed to evaluate the effect of progesterone. Because numerous other factors affect the development of SA, neutering may have only a minimal additive effect.

Age of acquisition, onset, and referral for behavioral examination—Age of acquisition was not a significant risk factor for separation-related behaviors in several studies.^9,14,30 Pierantoni et al³⁵ found an increase in behaviors, including attention-seeking behavior, that may portend the development of SA in dogs that were separated from their littermates (adopted) at 30 to 40 days of age, compared with dogs at 60 days of age. A confounder, however, is neurodevelopment that occurs at a young age, such that dogs should not be adopted before 8 weeks of age.¹¹ Dogs with SA arrived at their forever homes at an older age in 1 study,²⁸ but whether they already had SA or developed SA after arrival could not be determined.

Flannigan and Dodman⁹ found no association between the development of SA and age of acquisition. Approximately 50% of the dogs were acquired before 12 weeks of age, and no association existed between incidence of SA and adoption prior to 7 weeks of age. The comparative group was dogs with behavioral problems that did not include SA, rather than dogs without behavioral problems. Hence, the finding by Pierantoni et al³⁵ that behavior problems are more common in young adopted dogs could not be excluded as the reason for the findings in the Flannigan and Dodman study.⁹

Reliable data regarding the age of onset and presence of SA are lacking. Most studies^8,17,34 are retrospective and rely on dog-owner estimates of age of onset, with the latter subject to owner recall and therefore possibly inaccurate information. Owner-reported age of onset ranges from 13.3 to 51.8 months for dogs in a treatment study,⁸ although individual dog data are not provided. In another study of dogs with SA,³⁴ 21 dogs were 6 to 12 months of age, 54 dogs 1 to 5 years, 20 dogs 5 to 10 years and 4 dogs > 10 years, with an overall mean age of 21.6 months, and age was not associated with development of SA. In a prospective study,¹⁹ dogs exposed to various social and environmental circumstances (none that were intentionally scary) between 5 and 10 months of age were less likely to display signs of SA at 18 months.

Flannigan and Dodman⁹ also reported on whether a difference in the age at the time of referral for behavioral examination between dogs with and without SA was significant. The difference was not significant.

Previous ownership—Neither King et al³⁴ nor Flannigan and Dodman⁹ reported an association between previous ownership and signs of SA. Not all previously owned dogs were abandoned. Dogs that were relinquished through shelters or rescues or abandoned at veterinary hospitals are more commonly reported with SA in some studies.^9,23

Number of people and other pets in the household—Three studies^8,23,34 do not reveal an association between the number of household members and SA, but 1 study⁹ reveals that 1-person households are more likely to have dogs with SA, compared with other behavioral problems. Dogs that were referred to a behavior specialist for evaluation and lived in a 1-person household are approximately 2.5 times as likely to have SA. This finding suggests that multiperson households may mitigate the development of SA or that the same attention ability to attend to dogs' needs may not occur in multiperson households. Also, this finding may be attributable to the high burden of care, reduced social interventional opportunities, or owner distress owing to their dog being their only companion. With regard to other pets in the household, several studies do not reveal an association between the presence or absence of other pets in a household and SA.^{3,9,23,28,30,34}

Owner gender—Results of 3 studies^9,31,34 do not indicate an association between owner gender and SA. However, McGreevy and Masters³² found an increased probability of separation-related distress for dogs with an increased number of human females in the household.

Owner perception—Gonzalez-Ramirez et al³⁶ compared groups of owners whose dogs had or did not have SA. Each owner provided their perception of their dog's overall behavior. Dogs that did not have signs of SA had lived with their owners longer and had higher owner-rated trainability scores, and their owners had a lower perception of their own stress. Interestingly, this suggests that owner lifestyle and stress level may greatly affect their dog's behavior and subsequently owner perception of their dog's behaviors. An owner who has a high stress level may believe that their dog is burdensome or they have a stressful relationship with their dog, even believing that it is less trainable, regardless of whether this is true. A highly stressed relationship with their dog may contribute to its high level of anxiety and unwanted behaviors, such as SA.³⁶

Source—Bradshaw et al¹⁷ found no difference in the prevalence of SA between dogs acquired from a rescue organization and dogs acquired through breeders, but with the prevalence of SA in the breeders' dogs higher than expected. Palestrini et al²³ found that of 23 dogs with separation-related problems, 8 were acquired from an animal shelter and 4 each were acquired from a dog breeder, a pet store, or another person; 3 were stray dogs. McGreevy and Masters³² found a decreased risk of separation-related distress for dogs acquired from family and friends, compared with those acquired from pet shops.

Dogs acquired from animal shelters have often been reported^3,9,30 to be at high risk for developing SA, but whether a cause-and-effect relationship exists is unclear. A shelter environment has specific stress-ors that may predispose sheltered dogs to anxiety-induced conditions or worsen pre-existing clinical or subclinical anxiety-induced conditions. Stressors include exposure to high levels of noise, exposure to novel experiences, abrupt separation from previous attachment figures, prolonged confinement, exposure to a continuously changing population of dogs, and reduced interaction with conspecifics and people. Many of these stressors can have profound effects on puppies if they experience these stressors during their sensitive period of socialization.³⁷ Flannigan and Dodman⁹ found that dogs acquired from animal shelters, rescue organizations, or veterinary hospitals or dogs that were strays more commonly had SA (vs other behavioral diagnoses), compared with dogs acquired from other sources combined (dog breeder, family or friend, and pet store).

Dogs acquired from pet shops had a greater incidence of house soiling and separation-related behaviors, compared with dogs acquired from dog breeders, but this relationship was confounded by owner-related factors.³⁸ McMillan et al³⁹ found an increased risk of developing SA and higher odds of escape behaviors for dogs acquired from pet shops, compared with dogs acquired from breeders. Conversely, Flannigan and Dodman⁹ found that acquisition of a dog from a pet shop was not significantly associated with increased incidence of SA. The chain of at-risk situations from breeding of sire and dam through placement at a pet shop is complex and may include stressed dams (epigenetic effects); inbreeding (genetic effects); lack of adequate early care, exposure to relevant stimuli during sensitive periods, and nutrition; early weaning and separation from littermates; and stress of transport and placement at a pet store. For each of these situations, a cause-and-effect relationship is unclear; nonetheless, these high-anxiety situations may alter the brain neurochemistry of dogs,^12,16 such that some pet-store dogs may be at risk for developing SA.

Obedience classes—Dogs with SA were less likely to have attended an obedience class than dogs with other behavior problems, although the difference between groups was small.⁹ Class attendance is associated with a decreased risk of SA in nearly all studies^3,30,40,41; however, preadoption counseling targeted to prevent SA in dogs adopted from an animal shelter did not have an effect on preventing SA.⁴² Bennett and Rohlf⁴³ found a protective effect of obedience classes on SA, but the effect could not be directly attributed to the class. This finding may reflect the effect classes may have, but owners who attend classes may also expose their dogs to various social and environmental situations. As noted, dogs exposed to various normal social and environmental situations through 10 months of age were less likely to later have SA. Possibly, class attendance exposes owners to beneficial information about canine behavior or owners who attend classes have more time to spend with their dogs; research on these possibilities is lacking.

Boredom—Boredom is an often attributed and seldom defined reason for how dogs behave when left alone at home. Most popular definitions include a lack of interest. Dogs living in the enriched environment of most households are unlikely bored. If dogs were bored, abnormal behavior would be expected to be most intense after being left for a prolonged period of time. However, with true SA, the peak intensity of abnormal behaviors such as barking and destructive activities occurs shortly after an owner's departure.^21,23 Also, dogs with SA do not inappropriately urinate or defecate, destroy household items, salivate, etc, when their owners are present but not interacting with them. Both findings refute the speculation that dogs with SA are simply bored.

Exercise—One retrospective study²⁷ of a large number of dogs (n = 3,264) reveals that dogs with SA exercised less than those without SA. In many cases, exercise reduces generalized anxiety in dogs, so exercise may be not be specific mitigating factor for SA.⁴⁴ Lack of exercise may affect the development of SA but not the development of anxiety.

Body weight may be related to exercise. However, body weight is not associated with SA in 1 study³⁴ that included body weight was an independent variable.

Trauma and change in routine—Separation anxiety is more likely to develop in a dog that has experienced a change in its owner's work routine, an addition of a person to the household, a change in its owner's routine for leaving home, and a traumatic event during its owner's absence, such as flooding, fire, or abuse.^8,9 However, no specific data were identified in a literature search regarding the frequency and duration of owner absences. Relinquishment to a shelter may also be considered as a traumatic event, such that SA may develop; however, a dog instead may be relinquished because of SA. Early maternal environments and behaviors (pre-, peri- and post-parturient) can also be inadequate and equivalent to traumatic events with respect to stressors and nutrition. Tiira and Lohi²⁸ found that poor quality of maternal care contributes to generalized anxiety, possibly SA, through direct or epigenetic effects.

Cognitive bias—Cognitive bias is an outcome of an affective state or mood that affects decision making and judgment, meaning that an animal that has negative cognitive bias views a neutral stimulus as negative (neutral stimulus viewed with pessimism) and an animal that has positive cognitive bias views a neutral stimulus as positive (neutral stimulus viewed with optimism). Cognitive bias tests are increasingly used across species to assess emotional factors that influence problem-solving behavior, including those factors associated with welfare (eg, neglect and anxiety-related behavioral problems). Mendl et al⁴⁵ reported that shelter dogs with higher separation-based distress scores correlated with pessimistic-like behavior (negative cognitive bias) exhibited during a cognitive bias test. Because Mendl et al⁴⁵ only assessed the correlation between distress scores and negative cognitive bias, whether negative cognitive bias increases the risk of developing SA or bias is the result of SA is unclear, but this finding may impact treatment (behavior modification and medication) of affected dogs.

Spoiling activities—Spoiling is when a pet owner has an excessive emotional connection and praises or overindulgences their pet through various activities, such as allowing their pet to follow them, sleep on the bed or in the bedroom, sit on furniture with them or on their lap, receive gratuitous treats or attention, and receive attention when initiated by the pet rather than to receive attention when initiated by only the owner. These activities have been proposed to be the impetus for the development of SA when hyperattachment is a criterion of SA.⁸ King et al,³² Flannigan and Dodman,⁹ and Storengen and Lingaas³¹ specifically asked dog owners about spoiling activities as a contributor to SA with allowing the dog on the bed or feeding the dog at the dining table as their key behavioral indicators; no association was identified between these behaviors and SA. King et al³⁴ noted that by the end of their study (prospective, randomized, double-blind, placebo-controlled study of treatment with clomipramine of dogs with SA), signs of SA lessened following clomipramine treatment, but almost all dogs that had slept on the bed continued to do so, indicating that affected dogs may not perceive this activity to be related to their distress.

Temperament, personality, and breed—Temperament is generally defined as a pattern of early-appearing traits that are stable over an animal's lifetime. These temperament traits form the basis of personality, which is defined as an animal's characteristics that account for its general patterns of behaviors or behavioral responses.⁴⁶ Implied with these definitions is a degree of heritability, which may be separate from that of liability risk (ie, some aspects of personality may make a dog at high risk for possessing factors which may influence the development of SA, without a discrete genetic polymorphism actually conferring a neurobehavioral risk). Salononen et al¹⁵ reported that owners who described their dogs as having patterns of behavior that were impulsive and hyperactive also indicated that their dogs had SA. However, the online behavior questionnaire was limited in its exploration of SA, containing only 1 question pertaining to it, so the extent to which this is a meaningful correlation is unknown.

Bradshaw et al¹⁷ found a higher prevalence of SA at 18 months of age for 7 litters of Labrador Retrievers (57%) and Border Collies (35%), compared with the prevalence in a contemporaneous general dog population (29%), suggesting that breed personality may play a role in the development of SA.

Data that indicate whether 1 breed is overrepresented in a general population of dogs that have SA are seldom available. Yet, Storengen and Lingaas³¹ report that Cocker Spaniels and Dachshunds were overrepresented, compared with the relevant years of registration of the Norwegian Kennel Club. In a large study¹⁵ of Finnish pet dogs, Miniature Schnauzers, Lagotti Romagnoli, and Soft-Coated Wheaten Terriers commonly exhibited signs of SA. Kennel clubs could collect semiquantitative, objective data in annual health studies and track prevalence rate of SA. In the design of a study, inadvertent selection of one trait while attempting to select for another trait may be the result of a poorly defined and assessed trait or hidden factors in the variance-covariance matrix structure. Although mixed breed dogs have been shown to have a greater incidence of SA than purebred dogs in many studies,^{3,15,17,31,41} this was not the case in 1 study.⁹ However, the previous studies' results may be confounded by an overrepresentation of mixed breed dogs in an animal shelter.

Genes and genetic risk—Increasingly, data suggest heritability of genes that increase the risk for the development of SA. van Rooy et al⁴⁷ identified genetic polymorphisms that enhanced the risk of SA in Australian Golden Retrievers. In that study⁴⁷ which included 22 Golden Retrievers with SA, 18 without SA, and 15 intermediates (dogs with some mild signs that did not reach the diagnostic criteria), separation-related behavior scores were significantly increased for dogs that were homozygous for the CAA (nucleotides) haplotype at the dopamine drd2 locus, compared with those without the haplotype. Scores were decreased for dogs with the TAA haplotype, but the score difference was not significant. Nonsignificant increases in scores were also associated with the GAG haplotype at the arginine vasopressin avpr1a locus and nonsignificant decreases in scores with the GCG haplotype. Similar risk factors have been identified for noise reactivity and phobia in dogs.²⁹

Attachment—Both human and canine attachment type may play a role in the occurrence of SA. Attachment is often asserted to be associated with spoiling dogs and the development of SA. Kwan and Bain⁴⁸ define attachment as a multidimensional combination of the amount of time for various activities an owner spends with their pet, interest in and emotional closeness to their pet, knowledge of their pet's required care, and their behavioral responses to their pet. Although this definition lacks a psychological basis, Kwan and Bain⁴⁸ indicated that owners who felt less attached to their pets were likely to feel less attached when their pet had behavioral problems and to relinquish them to a shelter. Yet, whether a behavior problem or a feeling of poor attachment develops first is unknown.

Podberscek et al⁸ asserted that attachment problems, particularly hyper- or overattachment, lead to SA. Hyperattached dogs were defined as those that followed the owners throughout the house, wanted to sit near them, greeted them intensely after a separation event, demanded attention or affection, liked physical contact, and were attached to 1 person in the household. Separation anxiety was defined as owner-reported whining, barking, or howling; shaking or shivering; excessive salivation; sulkiness or depression; inappropriate urination or defecation; mouthing or nipping; and destruction of furniture, doors, or doorframes when owners were absent or when the dog was left alone to sleep at night. The behavior modification evaluated in this study⁸ was intended to break hyper- or overattachment that has been engrained and conditioned by the owners' behaviors.

Hyperattachment has generally been defined as a dog's tendency to follow the owner and always be in their presence. Authors of a retrospective case-control study⁹ of 200 dogs diagnosed with behavior problems other than SA and 200 dogs diagnosed with SA evaluated how hyperattachment affected dogs with SA. Sixty-four percent (128/200) of the dogs without SA and 83.8% (167/199) of the dogs with SA were characterized by their owners as having a moderate to severe owner-following behavior. Although more dogs with SA followed their owners, most dogs without SA (64%) also did so. If following of the owner indicates hyperattachment, hyperattachment is not a criterion for the diagnosis of SA because most dogs without SA also followed their owners and some dogs with SA (16.2%) did not follow their owners.

Flannigan and Dodman⁹ also examined associations between SA and other behaviors that have been targeted in SA behavior modification protocols. Increased intensity of greeting and signs of distress or anxiety when owners picked up keys or put on their coats (owner-departure cues) occurred twice as often for dogs with SA, compared with dogs without SA, but these behaviors also occurred in one-third of dogs without SA. Signs associated with high diagnostic sensitivity and specificity (vocalization, destruction, inappropriate urination or defecation, hypersalivation, vomiting, or diarrhea) were overwhelmingly reported for dogs with SA, not those with hyperattachment, supporting the belief that SA is a diagnosis independent of hyperattachment.

Bradshaw et al¹⁷ found that rather than increasing the likelihood of SA, the more time owners spend with their dogs at a young age (6 to 9 months of age) and the more time owners socialize their dogs to people, other dogs, and new environments (eg, meeting strangers and unfamiliar children, experiencing new environments and dogs, and attending puppy classes), the likelihood of developing of SA at 9 to 12 months of age decreased. Although breeds differed in their separation-related behavior scores, the results were significant for Labrador Retrievers and Border Collies.

Several studies^7,49 have used a variation of Ainsworth Strange Situation Test, a rubric used to assess mother-infant attachment, to assess dog owner-dog attachment. Mother-infant or more broadly parent-child attachment and dog owner-dog attachment are similar, given a dog's domestication and artificial selection for behavioral traits that promote socialization and attachment with people.^6,7

Prato-Previde et al⁷ proposed that a modified Ainsworth Strange Situation Test could determine whether an owner-dog bond was an affectional bond or an attachment. An affectional bond endures over time, involves a specific person, is emotionally important, and involves a person who prefers to maintain proximity and contact with their dog and who become distressed when involuntary separation from their dog occurs.⁷ Attachment is a type of affectional bond that also includes a secure-base effect, defined as when a child or animal can engage in activities without the parent or owner, respectively, present because they feel secure to do so knowing that they can return to their caregiver, the secure base. A dog with a more secure base should be more inclined to explore an unknown room and play with a stranger if the owner is present. Four categories of behavior—secure-base effect, proximity seeking, search behaviors, and comfort seeking—were evaluated for 38 dogs by Prato-Previde et al.⁷ The data support a strong affectional bond, but the data do not confirm the presence of attachment because the data were ambiguous about a secure-base effect.

In an attempt to better adapt the Ainsworth Strange Situation Test to conditions relevant to dogs, Parthasarathy and Crowell-Davis⁴⁹ hypothesized that dogs with SA would show different patterns of attachment, compared with dogs without SA, when alone with their owners during testing. Dogs with SA should spend more time in close contact or proximity with their owners, rather than exploring a novel facility. The authors also hypothesized that when dogs are loose in the house, dogs with SA would also spend more time in close contact or proximity with their owners than dogs without SA. Lastly, the authors expected a relationship existed for the duration and frequency of anxiety-related behaviors between when dogs were alone during testing and when they were crated at home. Video recordings and a detailed standardized questionnaire were used to assess the aforementioned hypotheses. The 31 dogs with SA did not differ from the 44 without SA in all measures of contact or proximity during the test. The amount of time dogs spent by the doors through which their owners exited did not differ between the groups. Lastly, for the crated dogs, anxiety-related behaviors during the test did not correlate with anxiety-related behaviors when a dog was crated. The results of this study⁴⁹ did not support the assertion that SA is caused by hyperattachment.

As hinted by Parthasarathy and Crowell-Davis,⁴⁹ part of the problem with the assertion that SA is caused by hyperattachment and spoiling is the lack of information on how dogs that do not have any behavioral problems respond to being left alone by their owners. Young and adult dogs without SA that are separated from their owners in an unfamiliar environment or by unfamiliar people usually show abnormal behaviors owing to a perceived threat and will attempt to regain proximity to their owners. Adult dogs also show separation-related distress when they are separated from other household dogs in similar circumstances.^50,51

In a study⁵² of 45 dogs (15 with SA and 30 without SA), the behaviors of dogs during a brief separation from their owners and the quality of their reunions were assessed. Dogs without SA spent more time near their owners' chairs after the owners left the room, whereas those with SA spent more time at the door through which the owners exited and remained active for longer, suggesting that an owner may not be a secure base for affected dogs. Furthermore, dogs with SA did not show more affection than dogs without SA when reunited with their owners. Instead, dogs with SA were more active, preferring to run instead of maintaining body contact or close proximity to their owners, and did not wag their tails faster. These data suggest that dogs with SA are not hyperattached to their owners and may not view their owners as a secure base and be calmed by their presence. Therefore, dogs with SA and their owners may have an insecure relationship.

Konok et al⁵³ examined the effects of style of owner attachment and of owner and dog personality on the development of SA. Owners of 1,508 dogs completed an online questionnaire that consisted of 5 sets of questions pertaining to demography, separation behavior of their dog, owner personality, their dog's personality, and owner attachment. An owner with a more avoidant attachment style was more likely to have a dog with SA. Konok et al⁵³ inferred that an owner's avoidant attachment style was in part the cause of their dog's SA because the owner was less responsive and less sensitive to their dog's needs, especially in situations in which their dog may be distressed. Therefore, owners may not provide a secure base for their dogs, as speculated in the previous study by Konok et al.⁵²

Owner attachment and caregiving style were further examined through results of an online questionnaire that was completed by 125 dog owners for 146 dogs.³³ Questionnaires focused on the owners' feelings and opinions of their dogs' separation and greeting behaviors, vocalization, and urinary, bowel, and destructive behaviors when dogs are separated from their owners. Owners with low to medium concern for their dogs' anxiety, despite the presence of signs that were consistent with marked anxiety, were more likely to have an avoidant attachment style and to be less responsive to their dogs' needs. Dogs with such owners had various signs, usually physiologic or behavioral. Difference in the type of signs displayed could be related to the type of attachment that the dog perceived, and dogs that displayed predominantly physiologic signs may be analogous to children with an insecure-avoidant type of attachment. Dogs with predominantly behavioral signs may be analogous to children with an insecure-anxious type of attachment. Regardless of whether the analogies to children are valid, both types of dogs showed less affectionate behavior with reunion with their owners, further suggesting poor attachment.

When SA is considered to be caused by hyperattachment, an owner is often given instructions to ignore their dog for 30 minutes prior to their departure and for at least the time it takes the dog to stop greeting and attending to them after their return home in an attempt to break the attachment.⁸ Mariti et al⁵⁴ examined the physiologic and behavioral responses of 10 dogs without SA during a 3-minute period of separation from their owners under 2 conditions: each owner calmly petted and talked to their dog before they left home, or they ignored (did not interact with) their dog before they left. Dogs displayed signs consistent with calmness for a significantly longer during the 3-minute period than they did when they were not petted. Additionally, dogs' heart rates significantly decreased after the 3-minute period following being petted prior to the owners' departures. Calmly signaling to and communicating with the dog by its owner prior to the owner's departure may provide the dog with useful information that the owner is leaving but will return. Lack of information may worsen anxiety and may affect the extent to which a dog can calm itself.^{54,55,56,57,58,59}

Teixeira and Hall⁵⁶ evaluated whether high arousal or excited departures and greetings lead to increased frequency of separation-related behavior in newly adopted shelter dogs. They hypothesized that dogs exposed to high arousal (vs low arousal) interactions would spend more time following their owners' departure by the door through which their owners exited, be more active and vocal when left alone again, and would have increased responses to high arousal interactions over 10 interactions. Findings did not support their hypotheses; arousal interactions were not associated with subsequent departure behaviors over the 10 interactions. In the survey component (part 2) of their study,⁵⁶ they found that owners of dogs with SA did not report more enthusiastic or high arousal reunions between their dogs and them. These data suggest that instructions to break the attachment and ignore the dog may, instead, convey to the dog unclear information about the owner's attachment style and deprive dogs of information that may be helpful to them.

In a review article¹ of the current knowledge of the causes of SA in dogs, Ogata concurred that evidence is inconclusive to support the presence of hyperattachment in dogs with SA. The aforementioned studies reveal that over- or hyperattachment is not a cause of SA, and instead SA may be in part associated with an insecure attachment between a dog and its owner. Insecurity may be partly associated with an owner's inability or lack of interest in responding to their dog, which may lead to a lack of temporal and predictive signals and information that interactions provide to the dog. This is an important distinction because dogs engage in referential signaling and use it to gain important information. That information or lack of that information is then used to modify their behaviors, but not always in ways their owners intended.^57,58,59 Dogs that are ignored are deprived of such contextual information.

In conclusion, 7 areas for which data were published were identified as follows: developmental, acquisition, environmental, owner-related, behavioral, genetic, and general. A review of these areas produced 20 putative factors that may convey increased risk for the development of SA. All findings were correlational. The data published in the peer-reviewed literature indicate:

• Evidence that spoiling dogs causes SA is not available.

• Sex, neuter status, gender of owners, presence of other pets in the household, number of people in the household, having had a previous owner, or boredom may not be factors associated with the development of SA.

• Exercise and attendance at obedience classes may be protective against the development of SA, but protection could be attributable to related latent factors, such as broad exposure to other animals and environments or enjoyment of exercise.

• Age at acquisition and early-age traumatic experiences, which may be related, warrant further research.

• Some evidence exists that at least in some breeds, genetic polymorphisms may increase the risk for the development of SA. Temperament and personality have not been sufficiently or consistently examined to assert their effect on the development of SA.

• Owner attachment style appears to play a larger role than expected in whether a dog develops SA, with owner attachment style likely related to an owner's perception of their dog's behavior.

• Shelter, rescue, and mixed breed dogs that have SA are overrepresented, but mixed breed dogs are overrepresented in animal shelters. Yet, owner attachment style and their perception that their dog's behavior is a problem place their dog at increased risk for relinquishment.

These findings are summarized in Table 2.

Separation anxiety is one of the most common behavioral conditions that affects dogs. Data are still lacking that may help with the development of an adequate risk assessment tool, mainly because studies have focused on the now recognized erroneous assumption that over- or hyperattachment is an important cause of SA. Recent research has shifted in focus from the belief that a dog with SA is hyperattached to its owner to the belief that a dog's behaviors are partly in response to owner attachment style and has continued to include investigations of possible genetic, epigenetic, and experiential risk factors for the development of SA.