Introduction

Adverse events can result from administration of any medication or product. These events can present clinically as a range of signs from mild (eg, lethargy or injection site soreness) to severe (eg, anaphylactic shock or death). Understandably, concerns about the possibility of more serious events can cause owners to decline important services that protect against severe and potentially fatal infectious diseases. Heartworm disease is a preventable disease caused by Dirofilaria immitis parasite infection that affects pets and other animals in the US and globally. If left untreated, the parasitic infection can lead to serious or fatal pulmonary and cardiovascular disease.¹ Prevention in the US is encouraged through consistent administration of a macrocyclic lactone product approved by the US FDA. While many approved products are oral or topical, injectable sustained-release moxidectin products (ProHeart 6 and 12) have been available in the US, with the 6-month product initially approved in 2001 and the 12-month product approved in 2019. Each injection is a suspension of moxidectin-containing microspheres that gradually release the parasiticide to prevent infection for the 6 or 12 months, with dosage based on dog weight. By providing 6 or 12 months of protection through parenteral (SC) administration, these products reduce noncompliance as a factor in failure of protection against heartworm disease.

Adverse events associated with injectable heartworm prevention products are reported rarely and include injection site reactions, vomiting, diarrhea, weight loss, listlessness, seizures, and elevated body temperatures.^2,3 For the 12-month product, adverse events observed during investigational field studies supporting FDA approval also included mild to moderate anaphylactoid reactions.³ On the basis of reports of adverse events to the manufacturer and the FDA for the 6-month product, the incidence of adverse events is estimated to be 3.9 events/10,000 doses sold between 2008 and 2016.⁴ However, the estimate is based on voluntary, passive reporting of adverse events by veterinary professionals and pet owners. Such a reporting system likely results in under-reporting of adverse events. Under-reporting may be due to factors such as low confidence in whether the adverse event was caused by the product, unawareness of how to report such events, inconsistent follow-through of reporting, and selection bias (eg, more serious events are more likely to be reported).⁵ Unfortunately, once products are launched, there is no benchmark for evaluating whether reports of adverse reactions are truly representative. Additionally, the actual number of doses administered (denominator data) is unknown. The only available estimate is made on the basis of product sales and a standardized dog weight.⁴ A higher-confidence estimate of adverse event risk would supplement the knowledge base and better support informed decision-making about product benefits and risks in the individual dog, including identifying potential risk factors and breed differences. Given current evidence that the true incidence is relatively low, a very large study population would be needed to provide a reliable estimate.

A large database of pet medical record information was available for retrospective query to generate a more accurate estimate of the incidence of uncommon or rare acute adverse events. An earlier study,⁶ conducted after concerns of serious adverse events following administration of the 6-month product, used a large pet medical record system. Since the 2008 reintroduction of the 6-month injectable product in the US, no study has been published of the incidence of acute adverse events related to the injectable products in client-owned dogs, nor have potential risk factors for these events been identified in canine patients. This study was conducted in a large population of client-owned dogs over a 5-year period to define the true incidence of adverse events more accurately and identify patient risk factors.

Methods

A retrospective study was conducted utilizing pet medical records of a large primary veterinary care network, currently consisting of over 1,000 primary care small animal hospitals located in 42 US states, the District of Columbia, and Puerto Rico. All hospitals in the network utilize the same proprietary pet medical record and invoicing (PMRI) system (PetWare; Mars Inc), with data uploaded nightly to a centralized data warehouse in Vancouver, Washington. The PMRI system contains both structured and unstructured fields. Structured fields (ie, list of options for selection) include physical exam observations, diagnosis, and invoice information on services provided and products administered and dispensed; unstructured fields include free-text information, such as detailed notes on physical examination findings and diagnostic and treatment plans.

All canine visits from January 1, 2016, through December 31, 2020, in which one of the injectable sustained-release moxidectin products was administered were extracted from the data warehouse. Patient visit information included the following: dog age, body weight, breed, sex/neuter status, and vaccinations administered (distemper, adenovirus, parainfluenza, and parvo; leptospirosis [4-way]; Bordetella [+/− parainfluenza]; rabies; Lyme; and bivalent canine influenza) during the visit. To remove confounding due to vaccination,⁶ dog visits that included administration of any vaccine(s) were excluded from the dataset.

An acute adverse event was defined as having met at least one of the following criteria within 3 days of administration of the moxidectin product:

• Diagnosis of allergic reaction, acute allergic reaction, urticaria, anaphylaxis, immune-mediated disease, vaccine reaction (mild/localized, moderate systemic, severe systemic, or unspecified severity), dead on arrival, and death while under treatment.

• Formal documentation of adverse reaction, a structured item in the PMRI system, which includes recording of any of the following 4 reactions: vomiting/diarrhea, swelling/hives/mass, anaphylaxis, or “other clinical signs.”

The postadministration window of 3 days was based on previous research reporting that over 95% of allergic reactions associated with the moxidectin product occurred in the first 3 days following exposure.⁶

Because the list of diagnoses available for selection in the PMRI is long and there are no criteria for selection of diagnoses, the diagnosis selected to any patient visit may have been affected by awareness of existing options in the drop-down list by the veterinary team member record. Thus, some adverse events associated with the moxidectin product may have had a diagnosis of vaccine reaction even though no vaccine had been administered on that visit. To capture as many adverse events as possible, vaccine reaction-related diagnoses were included.

Patient risk factors included in the analyses are characteristics consistently recorded in patient records, as follows: sex/neuter status, breed, and age and weight at visit. Dog visit ages were grouped into 8 categories: 0 to 1.5 years, > 1.5 to 2.5 years, > 2.5 to 3.5 years, > 3.5 to 5.5 years, > 5.5 to 7.5 years, > 7.5 to 9.5 years, > 9.5 to 11.5 years, and > 11.5 years. Visit weights were grouped into 10 categories: 0 to 5 kg, > 5 to 10 kg, > 10 to 15 kg, > 15 to 20 kg, > 20 to 25 kg, > 25 to 30 kg, > 30 to 35 kg, > 35 to 40 kg, > 40 to 45 kg, and > 45 kg.

All analyses were performed using SAS version 9.4 (SAS Institute Inc). Incidence of an acute adverse event was estimated by determining the number of patient visits in which a dog met the case definition of an acute adverse event divided by the total number of doses of the heartworm preventive administered. Initial descriptive statistics were performed including estimating the incidence by breed. Breeds included those that had at least 1,000 unique dogs of that breed that received the injectable heartworm preventive product. Those breeds with an incidence greater than that of the mixed-breed dogs, along with visit age, sex/neuter status, and visit weight were included in univariate analyses to identify those significantly (α = 0.05) associated with an acute adverse event, a binary outcome variable. Only the breeds and other risk factors found to be statistically significant on univariate analysis were included for the multivariable analysis. Because the product can be administered more than once to a dog during the 5-year time frame, a mixed-effects logistic regression model was utilized for the univariate and multivariable analyses, with the pet patient ID number as the random effect. Crude and adjusted ORs and 95% CIs were generated from the output of the regression models, and statistical significance was evaluated using α = 0.05.

Results

From January 1, 2016, through December 31, 2020, there were 1,119 hospitals in the primary care veterinary hospital network, and 5,015,678 doses of the injectable heartworm preventive products were administered to canine patients. Removing those visits during which a vaccine was also administered, 1,399,289 (27.9%) doses remained in the dataset. Due to the 12-month product not becoming available in the US until mid-2019 (year 4 of the study period), approximately 97.5% (1,363,885) doses administered were the 6-month product. Due to this disparity in representation, differentiation between receiving the 6- and 12-month product was not made for the analyses.

The doses were administered to 694,030 unique dogs, with approximately 52.3% (363,214) being male and 89.5% (621,434) spayed or neutered. The range of doses received per dog in the 5-year study period was from 1 to 11 (average, 2.0/dog), with approximately half of the dogs (350,249 [50.5%]) having received only 1 dose, while 158,534 (22.8%) received 2 doses, 87,499 (12.6%) received 3 doses, and 97,748 (14.1%) received at least 4 doses. Ten dogs received 11 doses of the injectable preventive during the 5-year period.

Among the 1,399,289 doses administered, 1,999 visits met the case definition of an acute adverse reaction, allowing an estimate of 14.3/10,000 doses administered. This equated to 1,875 dogs that had at least 1 visit with an acute adverse event (or 27.0 dogs with at least 1 event/10,000 unique dogs). The most frequently recorded diagnoses were nonspecific allergic reaction (587 [29.4%]), mild vaccine reaction (386 [19.3%]), acute allergic reaction (207 [10.4%]), and moderate vaccine reaction (191 [9.6%]). Urticaria was diagnosed in 179 (9.0%) cases, including 16 dogs that received one of the allergic or vaccine reaction diagnoses. More severe events had lower occurrence, including 53 (2.7%) cases of severe vaccine reaction, 2 (0.1%) cardiac arrest, 2 (0.1%) death while under treatment, and 1 (0.05%) dead on arrival. In total, among the 1,999 events identified, 283 (14.2%) recorded formal documentation of adverse reaction, with all recording vomiting/diarrhea, swelling/hives/mass, anaphylaxis, or “other clinical signs.”

Descriptive statistics for the study dataset are presented (Tables 1 and 2). The annual doses administered each year were consistent across years, with numeric decreases each year in the incidence of an acute adverse event. Slightly more doses were administered to male dogs (intact or neutered) than females, with the lowest incidence of an adverse event in female spayed dogs. Those dogs 1.5 years old or younger and those older than 11.5 years each made up smaller proportions of recipients, while the 3.5- to 5.5-year age group was the highest proportion. Remarkably, the incidence of an acute adverse event was highest (47.5/10,000) in dogs ≤ 1.5 years, was next highest in dogs 1.5 to 2.5 years (25.4/10,000), and decreased from 15.3/10,000 in dogs 2.5 to 3.5 years to 6.4/10,000 in dogs > 11 years. More doses (55%) were administered to small or medium-sized dogs (up to 10 kg) with declining administration in larger dogs (6.2% over 40 kg). Dogs with visit body weight between 15 and 35 kg had rates > 15.1 events/10,000 doses, whereas lighter and heavier dogs had slightly lower incidence (10 to 13.8 events/10,000 doses). Those 17 breeds with incidence of an acute adverse event greater than that of the mixed-breed dog are presented (Table 2; approx 16.0 events/10,000 doses administered). Pit Bulls and American Staffordshire Terriers had the highest incidence with over 45 events/10,000 doses. A list of other breeds with incidence of acute adverse event approximately equal to or less than that of the mixed-breed dog is available elsewhere (Supplementary Table S1).

The results of the mixed-effects logistic regression models are presented (Table 3). The univariate analyses found sex/neuter status, visit age, and breed to be significant risk factors. When compared to intact female dogs, both spayed female (P = .011) and neutered male (P < .001) dogs had lower odds of an acute adverse event; intact males were not significantly different (P = .806) from intact female dogs. When compared to dogs 1.5 years of age or younger, the odds of an acute adverse event decreased with age, and all age categories (> 1.5 years old) were significantly different (P < .001) from this youngest age group. Only those breeds with significantly greater odds of an acute adverse event compared to the mixed-breed dog (Pit Bull, American Staffordshire Terrier, French Bulldog, Rhodesian Ridgeback, American Bulldog, Boxer, and Boston Terrier) were carried forward into the multivariable model. The “other breeds” category including those 10 breeds not found to be statistically significant, as well as dogs of any other breed that received the injectable product, were compared against the mixed breed. The “other breeds” included 595,864 unique dogs that received 1,212,776 doses, with 1,399 acute adverse events (estimated incidence of 11.5 events/10,000 doses administered). Thus, the multivariable model included the breed categories as well as the 2 following risk factors found to be statistically significant on univariate analysis: sex/neuter status and visit age.

The multivariable model found that when controlling for the other variables, sex/neuter status (P ≥ .128 across the categories) was not associated with an acute adverse event. Visit age was still strongly associated with an event with the odds of an event decreasing with age after the dog was 1.5 years old (P < .001). Similarly, the included higher-risk breeds were still found to have significantly higher odds of an acute adverse event when compared to mixed breeds (P ≤ .007 across the breeds), while the group of “other breeds” was not significantly different from the mixed breed (P = .072).

Discussion

To the authors’ knowledge, this was the first large-scale retrospective study since 2005 to estimate the incidence of adverse events occurring within 3 days of administration of the extended-release injectable moxidectin product where the number of doses administered was known, and the first identification of canine risk factors that are associated with such events. The study supported that such events are uncommon in the general canine population in the US. The identified patient characteristics that are significantly associated with such an event include visit age and breed, which are similar to identified risk factors for vaccine-associated adverse events (VAAEs), with older dogs having decreasing risk compared to younger dogs.⁷ Compared to mixed-breed dogs, the higher-risk breeds that were identified through the multivariable analysis include the following: Pit Bulls, American Staffordshire Terriers, French Bulldogs, Rhodesian Ridgebacks, American Bulldogs, Boxers, and Boston Terriers. The list of higher-risk breeds identified in this study may not encompass all higher-risk breeds, as the analysis focused on more popular breeds (at least 1,000 unique dogs of that breed received the product), and analysis of those less popular breeds may identify further breed differences at risk of an event. As with any study suggesting breed differences, the question arises about the role of genetics in adverse reactions, particularly given some of these higher-risk breeds are also identified at higher risk for a vaccine-associate adverse event.⁷ Unlike the vaccine-related adverse events, body weight was not found to be a significant risk factor, and this may be related to the weight-based dosing of the injectable heartworm product.

As with any retrospective study, there were several limitations to this study. The study focused on association and did not prove causality. Because dogs may return to primary care hospitals in the network and meet the case definition for an acute adverse event for other reasons (eg, comorbidities, receipt of other services or products on the same visit as heartworm preventive injection, bee sting after product administration, and death due to postdischarge traumatic event), we were unable to determine whether all 1,999 events identified were due to receipt of the product. Administration of inappropriate dose of the moxidectin product contributing to dogs meeting the case definition cannot be ruled out. Given the safety profile of the products,^2,3 this is not likely but not impossible. Due to the size and scale of this study, manual review of patient medical records was outside the scope of this study. In addition, there was the possibility of detection, selection, and reporting bias. Milder, localized events may be less likely to be detected in some dogs with certain body characteristics, such as those that are highly muscular or have longer coats or thicker undercoats. Dogs that experienced an adverse event after receipt of the product at a younger age might not continue to receive the product for heartworm prevention as they get older, thus possibly explaining the decreasing odds of an acute event with increasing visit age. Some dog owners may be less likely to opt for an injectable heartworm product for their dog. Vaccine history nor history of VAAE was not controlled for in the study design or analysis. It is possible that dog owners highly concerned about VAAE and dogs with history of VAAE are less likely to receive an injectable heartworm product when oral and topical options are available. In addition, dogs with history of VAAE may be more likely to be marked with a diagnosis of vaccine reaction when any clinical signs are seen after administration. It is possible that some of the dogs receiving the injectable preventive were pretreated with diphenhydramine either at home or in clinic out of an abundance of caution of the owner, and this was not controlled for in the study. With the inclusion of vaccine reaction in the case definition for an acute event, this may have included dogs that did not have a reaction to the moxidectin product. Some records of patients that were known to have had a vaccine reaction had a vaccine reaction–related diagnosis on every visit, regardless of visit reason and services provided or occurrence of adverse event during the visit. This may be done to facilitate identification of a pet with VAAE history without having to review all previous medical visit notes during a subsequent vaccination visit. On the other side of the reporting bias spectrum, pets that did not return for a follow-up visit for a possible adverse event or whose diagnosis was not recorded in the structured diagnosis field of the PMRI within 3 days of product administration would not have been captured as events. An example of a missed event would be a dog developing mild signs but not presented for a follow-up visit due to spontaneous resolution. Another possibly missed event is a dog that developed severe adverse effects but was taken directly to a veterinary emergency service and did not return to a network hospital within 3 days of product administration.

Finally, the breeds recorded in pet medical records are generally owner reported rather than confirmed by breed registry papers or genetic analysis. Thus, breed misclassification cannot be ruled out, but given the size of the study population, the extent of potential bias on breed-level incidence estimates is questionable.

Comparison of this study’s incidence estimate with that previously published in 2005 would not be appropriate to draw conclusions about changes in safety, as there are differences in the studies’ objectives and the methodologies used to evaluate frequency.⁶ Unlike the current study, the 2005 study⁶ included both immediate and longer-term outcomes associated with the injectable product. In addition, this study’s incidence estimate should not be compared to that provided by the manufacturer,⁴ as far as estimated accuracy or quality. While this study’s incidence estimate is greater than that shared by the manufacturer, this may be at least partly explained by the differing approaches and data sources for these estimates. Each approach has different assumptions affecting the interpretation of the incidence estimates. The evidence provided by these studies supports that adverse events that occur after injectable heartworm product administration appear to be uncommon. Knowing this and the risk factors identified in this study can support veterinary team recommendations and owner decisions about whether the injectable moxidectin products fit the individual dog’s needs on the basis of their characteristics and health history, as well as owner expectations and accepted level of risk.

Supplementary Materials

Supplementary materials are posted online at the journal website: avmajournals.avma.org