The effect of time until surgical intervention on survival in dogs with secondary septic peritonitis

Abstract

This retrospective study examined the effect of time to intervention on outcome in cases of dogs with secondary septic peritonitis, and also searched for other potential prognostic factors. The medical records of 55 dogs were reviewed. No association was found between outcome and the time from hospital admission to surgical source control. However, several other factors were found to influence survival, including: age, needing vasopressors, lactate, pre-operative packed cell volume, serum alkaline phosphatase, serum total bilirubin, and post-operative serum albumin. These values were then used to create accurate pre- and post-operative survival prediction models.

Introduction

Secondary septic peritonitis is the most common form of peritonitis encountered in companion animals and results from bacterial contamination of the abdomen, from internal fluid leakage, or external penetration (1). Leakage from the gastrointestinal (GI) tract is the most common cause of septic peritonitis in dogs and cats, but other sources include the biliary or urogenital system (2). Treatment of secondary sepsis is focused on hemodynamic stabilization of the patient, antibiotic administration, and surgical source control and decontamination (2,3).

Septic peritonitis is a surgical emergency and the condition is associated with a guarded to poor prognosis, with most studies reporting survival rates of approximately 50%, but ranging from 36.4% to 85% (3,4). As these patients require intensive treatment, experience high complications rates, and have prolonged hospitalization, it is important to be able to appropriately guide owners’ expectations regarding outcome, length of stay, and possible complications.

Underscoring the difficulty in predicting outcome in these cases, several studies have shown no difference between survivors and non-survivors with regard to signalment, pre-operative clinicopathologic variables, or location of contamination (3,5,6). Others have examined the effects of various treatments on outcome including early enteral nutrition, open versus closed abdominal drainage, early antibiotic administration, and the use of canine albumin transfusion; however, no clear survival advantage has been demonstrated (3,7–9).

While human hospitals have recently documented a clearer correlation between early diagnosis and treatment of sepsis, the relationship between time and outcome in veterinary patients with septic peritonitis is less clear. To the authors’ knowledge, there is only one previous veterinary study that investigated time until intervention (source control) and its impact on mortality (10).

The primary goal of our study was to determine if there is an association between time from admission to surgical intervention (TTI) and survival in patients with septic peritonitis. Our hypothesis was that patients which had a shorter TTI would have higher survival rates. A secondary goal of the study was to identify potential prognostic indicators of survival in patients with septic peritonitis.

Materials and methods

The medical records of all dogs diagnosed with septic peritonitis that underwent surgical intervention at Veterinary Specialists & Emergency Services (VSES) between May 2007 and April 2015 were reviewed. Cases were included based upon reported intra-operative findings (gross leakage of viscus contents or ruptured abscess.) Cytology (presence of bacteria in abdominal effusion) and culture were used to further support the diagnosis if exploratory surgery was suggestive but not definitive of secondary septic peritonitis.

Surgery was performed by a board-certified surgeon, a surgical resident supervised by a board-certified surgeon, or a surgical resident alone. All dogs were treated with the objective of source control (drainage and debridement of the infectious nidus as well as restoration of optimal function to that area), and administration of broad-spectrum antimicrobials. Post-operative abdominal drainage was performed in some cases.

Data collected included age, breed, gender, body weight, duration of hospitalization prior to surgery, duration of clinical signs prior to surgery, history of recent abdominal surgery or other invasive procedures, recent administration of steroids or nonsteroidal anti-inflammatory drugs (NSAIDs), source of contamination, nature of the abdominal effusion, bacterial culture and sensitivity (if available), type of surgery performed, level of surgeon’s experience, length of surgery, placement of a feeding tube, placement of an abdominal drain, whether or not patients were treated for hypotension pre-, intra-, or post-surgery, pre-operative coagulation status (normal versus prolonged), administration of a transfusion, whether or not the patient was evaluated by their primary veterinarian prior to admission, as well as pre-operative lactate, alanine aminotransferase (ALT), blood urea nitrogen (BUN), creatinine, albumin, hematocrit, total protein, potassium, white blood (cell) count (WBC), and post-operative albumin.

When available, hemograms were noted for the presence of bands or toxic change to the neutrophils. If a patient had surgery for septic peritonitis more than once at this hospital, only the first encounter was defined as the event. For patients that survived surgery, data for post-operative packed cell volume (PCV), total solids (TS), and lowest recorded post-operative serum albumin level were collected.

Final outcome was categorized as survived, died/arrested, or euthanized intra- or post-surgery. Euthanasia was performed during surgery if the owners did not wish to proceed after surgical exploration, based on perceived poor quality of life or increased financial estimate. For our study we classified patients as survivors or non-survivors, grouping patients euthanized in surgery with patients which were euthanized or died after surgery to assess final outcome. We combined cases of dogs that were euthanized after surgery with those that arrested after surgery, because these patients were usually euthanized due to perceived imminent death, in an effort to palliate pain and suffering.

Statistical analysis

To compare survivors to non-survivors, Fisher’s exact test (2 × 2 tables) or Chi-square test was used for categorical variables. A 1-way analysis of variance (ANOVA) was used to compare continuous variables. Categorical data are presented as frequencies and percentages, and continuous data are presented as means ± standard deviation (SD). Odds ratios were calculated when applicable.

After identifying variables with significant differences between groups, a logistic regression was used to create a predictive model of survival using selected variables. All analyses were performed using commercially available statistical software (IBM SPSS Statistics 20; IBM, Armonk, New York). Statistical significance was set at P < 0.05.

Results

Fifty-five cases met the inclusion criteria. In 44 cases diagnosis was made on the basis of surgical exploration, 8 cases were confirmed based on bacteria in the effusion and 3 were confirmed with abdominal fluid culture. There were 5 intact females, 5 intact males, 19 spayed females, and 26 neutered males. The ages of dogs ranged from 2 mo to 15 y, with the mean age of all dogs being 7.2 ± 4.25 y. The mean age of dogs that survived was 5.8 ± 3.9 y compared to the mean age of dogs that did not survive, which was 9.3 ± 3.7 y (P = 0.013). The overall survival in our population was 33/55 (60%). Six dogs died during surgery or after surgery. Nine dogs were euthanized during surgery, and 7 dogs were euthanized after surgery.

There was no association between location of leakage, or type of surgery and outcome (Table 1). The gastrointestinal (GI) tract was the most common source of leakage, which was caused by a perforating foreign body (n = 15), perforation of an ulcer (after NSAID or steroid use) (n = 9), dehiscence of a previous surgery site (n = 8), perforation of a neoplasm (n = 5), mesenteric torsion (with necrosis and leakage) (n = 1), gastric dilatation and volvulus (GDV) (from gastric necrosis) (n = 1), iatrogenic (bowel laceration from trocarization) (n = 1), gunshot wound (n = 1), and perforations of unknown cause (n = 5). The type of surgery performed did not have a significant association with outcome (Table 2).

Table 1.

Sources of abdominal contamination

Source of contamination	Survivors	Non-survivors
Stomach	8	5
Duodenum	9	5
Jejunum/ileum	8	7
Colon	1	1
Biliary/liver	1	3
Urogenital (Uterus/prostate/bladder)	6	1

Table 2.

Types of surgery performed

Type of surgery	Survivors	Non-survivors
Intestinal resection and anastomosis or Billroth I	17	9
Pyloroplasty, ulcer debridement with primary closure, enterotomy, or gastrotomy	9	2
Pyometrectomy	2	1
Cystotomy	2	0
Abscess debridement	2	0
Cholecystectomy	1	1

The time from admission to surgery was analyzed on a linear basis and also categorically by classifying patients in 6- or 12-hour groups. The 6-hour time blocks represented a reasonable and recommended time period for any medical therapy or preoperative stabilization to take place, and reflected the urgency with which the case was approached. For either method of analysis there was no correlation between time from admission to surgery and outcome.

The length of surgery ranged from 10 min to 5.5 h. The average length of surgery was 2.53 ± 1.03 h when performed by a resident alone, 1.63 (± 0.64) h when performed by a diplomate, and 1.89 ± 1.16 hours when performed by a resident supervised by a diplomate.

We also examined the length of surgery with animals which were euthanized in surgery removed from the data set. The mean length of surgery was 2.36 ± 1.05 h. Surgery was longer when residents were alone in the operating room (2.86 ± 0.88 h) than when either a diplomate alone (1.63 ± 0.64 h) or a diplomate and a resident (2.12 ± 1.14 h) were present during the procedure (P = 0.007). We did not find this when all cases were analyzed. Overall length of surgery though, was not statistically different when those dogs which had been euthanized during surgery were removed from the data set.

Results for univariate analysis of pre- and post-operative laboratory and patient variables are listed in Table 3. The relationship between albumin level (pre- and post-operative) was examined and as a continuous variable was not significantly associated with survival. When we designated a cutoff point of 10 g/L to constitute severe hypoalbuminemia, post-operative severe hypoalbuminemia was significantly associated with non-survival [χ² = 7.18; P = 0.01 (OR: 7.33)].

Table 3.

Patient and laboratory values

Parameter	n	Survivors	n	Non-survivors	n	P
Lactatea (mmol/L)	50	2.14 ± 1.12	30	4.93 ± 4.01	20	0.001
Post-operative albumin (g/L)	38	12.2 ± 4.2	27	9.4 ± 3.7	11	0.061
Agea (years)	55	5.8 ± 3.87	33	9.25 ± 3.71	22	0.002
Hospitalization (days)	45	5.85 ± 2.54	33	3.5 ± 2.07	12	0.006
Time from admission to surgery (h)	55	11.42 ± 9.63	33	14.86 ± 12.94	22	0.264
Length of surgery (h)	55	2.42 ± 1.05	33	1.83 ± 1.15	22	0.055
Duration of symptoms (d)	55	3.99 ± 6.74	33	3.74 ± 4.2	22	0.873
ALT (U/L)	55	80.82 ± 137.35	33	141.41 ± 202.27	22	0.191
BUN (mmol/L)	54	6.20 ± 4.25	33	8.76 ± 5.53	21	0.06
Creatinine (μmol/L)	53	104.31 ± 61.00	32	129.95 ± 99.89	21	0.254
Pre-operative albumin (g/L)	53	23.3 ± 6.5	32	20.7 ± 7.3	21	0.179
Pre-operative PCVa (%)	53	45.94 ± 10.08	32	39.41 ± 10.2	21	0.026
Pre-operative total protein (g/L)	55	54.4 ± 13.2	33	55.6 ± 15.4	22	0.75
K (mmol/L)	53	4.18 ± 0.85	31	4.1 ± 0.89	22	0.751
WBC × 1000 (/mm3)	55	13.19 ± 6.43	33	17.45 ± 13.77	22	0.128

^aIndicates variable with significant difference between survivors and non-survivors.

ALT — alanine aminotransferase; BUN — blood urea nitrogen; K — potassium; WBC — white blood cells.

Mean pre-operative serum lactate levels in survivors were significantly lower (2.11 ± 1.10 mmol/L) than those of patients that died or were euthanized (4.71 ± 4.02 mmol/L; P < 0.05). Using a cutoff of 3 mmol/L was a strong predictor of survival (OR: 0.103, P = 0.001, χ² = 2.01).

Pre-operative bilirubin and serum alkaline phosphatase (ALP) levels were significantly higher in those dogs which died versus those which survived (P = 0.04 and P = 0.001, respectively) and higher in those which died a natural death than those which were euthanized (P = 0.05 and P = 0.006, respectively) but not significantly different between those which survived and those which were euthanized (P = NS) (Table 4).

Table 4.

Differences in liver values between patients which survived, died, or were euthanized

Parameter	Survived	Died	Euthanized
Pre-operative total serum bilirubin (μmol/L)	7.87 ± 13.00	32.16 ± 56.44	7.01 ± 10.26
Pre-operative serum ALP (U/L)	262.12 ± 355.2	1440.17 ± 1991.9	348.13 ± 333.39

ALP — alkaline phosphatase.

Twenty-nine of 55 (53%) patients suffered from hypotension (SYS < 90 mmHg, MAP < 60 mmHg) during hospitalization (pre-, intra-, or post-surgery). Of these patients, 13 (45%) received a vasopressor agent either pre-, intra-, or post-surgery. Fourteen patients had hypotension that occurred pre- or post- surgery (i.e., not while under anesthesia). When excluding those patients in whom hypotension only occurred during surgery, 6 patients with hypotension received a vasopressor.

The presence of hypotension during hospitalization or surgery was not associated with survival (P = NS). This remained the case when patients in whom hypotension only occurred during surgery were excluded. Patients which were hypotensive and received a vasopressor had a significantly higher rate of death or euthanasia [P = 0.023 (OR: 5.02)]; however, this difference was not significant when examining only the cases in which vasopressor use occurred outside of anesthesia.

The use of vasopressors was also significantly (P = 0.010) correlated with pre-operative lactate level. Patients which received vasopressors had mean pre-operative lactate levels of 4.84 ± 4.22 mmol/L, compared to 2.53 ± 2.06 mmol/L in patients which did not receive vasopressors. When examining only cases in which vasopressor use occurred outside of anesthesia, the lactate of those which received vasopressors and those which did not were 6.58 ± 4.56 mmol/L and 2.89 ± 2.57 mmol/L, respectively (P = 0.007).

Two models were created based on findings from the univariate analysis: 1 for predicting survival before surgery and 1 for post-operative survival. Both models accurately predicted outcome > 80% of the time, based upon our data. The model variables for pre-operative survival included age, PCV, pre-operative total bilirubin, serum ALP, and serum lactate level. A test of the full model was statistically significant indicating that the chosen variables as a group reliably distinguished between those dogs that survived versus those which died or were euthanized [χ² = 27.4, df (5); P < 0.001]. The Nagelkerke R2 of 0.58 indicated a moderately strong relationship between the prediction and grouping. The overall correct prediction for the model was 81.6% (70% for non-survival and 89.7% for survival). The overall odds ratio for the complete model is 1.45. The Wald criterion demonstrated that only age and lactate made significant contributions to the prediction (P = 0.018 and P = 0.005, respectively). For every additional year of age, the odds of survival decreased by 0.7. For every 1 mmol/L increase in serum lactate level the odds of survival decreased by 0.5. While pre-operative bilirubin, ALP and PCV as individual variables did not contribute significantly to the model, when these variables were removed from the model development, the overall correct prediction decreased to only 75.5% with a decrease in survival prediction to 79.3%.

The strongest post-operative prediction model included those variables from the pre-operative prediction model (age, PCV, pre-operative bilirubin, ALP, and lactate) along with post-operative serum albumin level as a categorical variable (<1 versus ≥ 1) [χ² = 25.1, df (6); P < 0.001]. The Nagelkerke R2 of 0.72 indicated a moderately strong relationship between the prediction and grouping. The overall correct prediction for the model was 85.7% (90.9% for non-survival and 83.3% for survival). The overall odds ratio for the complete model is 2.18. The Wald criterion demonstrated that only lactate made a significant contribution to the prediction (P = 0.05) with a trend for post-operative albumin (P = 0.06). While age, pre-operative bilirubin, ALP, and PCV as individual variables did not contribute significantly to the model, when these variables were removed from the model development, the overall correct prediction decreased to only 80% with a decrease in death/euthanized prediction to 72.7%.

Discussion

The results of our study did not support our hypothesis that shorter TTI impacted outcome. Factors that were significant independently and as part of a predictive model were age, requiring vasopressors, lactate, pre-operative PCV, serum ALP, serum total bilirubin, and post-operative serum albumin.

A previous study of septic peritonitis found that survivors had significantly lower pre-operative serum values for the liver enzymes alanine aminotransferase and γ-glutamyl transferase than did non-survivors (11). While our study found similar results, these should be interpreted cautiously. There were 2 patients in the group that died which had biliary tract obstruction and consequently severely elevated ALP and bilirubin, affecting the mean. Packed cell volume has also been associated with prognosis in animals and humans with critical illness and sepsis (12). However, PCV can vary due to dehydration or IV fluid dilution, and although the difference was significant, the mean PCV of non-survivors was still within the reference range. While not likely to be useful on their own, initial PCV, serum ALP, and total bilirubin may be useful components of a prognostic or risk model for patients with septic peritonitis, as they could be sentinels of systemic illness.

Increased patient age was negatively associated with survival, which, to the authors’ knowledge, has not been reported for septic peritonitis in animals. While increased age has been associated with a poorer prognosis in humans with abdominal sepsis (13,14) there is also a strong chance that, within our population, a euthanasia bias played a role in the significance of age. While a frank discussion regarding costs and prognosis is always undertaken with an owner prior to surgery, owners may view these differently depending upon the age of their pet.

Pre-operative serum lactate of > 3 mmol/L was associated with a higher mortality rate. Lactate is an end product of anaerobic glycolysis, and increased plasma lactate is seen with inadequate tissue perfusion and subsequent hypoxia (15). In assessing dogs with GDV, one study showed that mortality was higher in patients with a lactate of > 9.0 mmol/L, and another showed that an initial lactate cutoff of 7.4 mmol/L was accurate for predicting gastric necrosis and outcome (16,17). Cortellini et al (18) found that in patients with septic peritonitis, survivors had lower mean initial lactate concentration and an inability to normalize the hyperlactatemia was associated with a poorer prognosis.

Pre-operative hypoalbuminemia is associated with increased risk of anastomotic leakage, and development of post-operative septic peritonitis (19,20). Furthermore, low serum albumin levels have been predictive of non-survival in both human and veterinary studies of abdominal sepsis and other critical illnesses (10,20,21). Goldwasser et al (21) found a 24% to 56% increase in the estimated odds of death for every 2.5 g/L decrease in serum albumin concentration, and concluded that serum albumin level was a highly sensitive indicator of preclinical disease and disease severity, and potentially had direct protective mechanisms. This association may be due to albumin’s role in maintaining intravascular volume through colloid osmotic pressure (COP), providing drug and hormone binding capacity, protection from oxidative damage, blood pH buffering, and mediation of coagulation. While albumin is not a specific component used in wound repair, hypoalbuminemia is associated with delayed wound healing (22).

Prior studies of septic peritonitis in dogs that found no association between albumin and outcome examined pre-operative, but not post-operative serum albumin values (5,11,23). Our study found an association between post-operative serum albumin and survival, similar to that found by Craft and Powell (8). Albumin can be lost at an accelerated rate in the septic patient via leakage through endothelial and peritoneal membranes, which can be compounded by the dilution effects of aggressive crystalloid therapy (22). Thus, the post-operative value may represent their “true” serum albumin level and be of more value in predicting survival and guiding transfusion after surgery.

Many factors related to abdominal surgery can influence blood pressure, including anesthesia-induced vasodilation and myocardial depression, acute hemorrhage, hypothermia from an open abdomen, obstruction of venous return secondary to positive pressure ventilation, dorsal recumbency, and packing or retraction of organs to isolate a viscus (24). These factors may have confounded the true relationship between blood pressure and survival in our study. When examining cases in which the hypotension only occurred outside of surgery and anesthesia, there was no association between hypotension or vasopressor use and survival.

Cases which required vasopressors likely represent patients in more severe or even refractory shock, where physiologic compensation mechanisms and intravascular volume restoration fail to correct hypotension, as is often seen with sepsis and systemic inflammatory response syndrome. Circulating inflammatory hormones can have vasodilatory effects on blood vessels, causing NO-mediated vascular smooth muscle relaxation, and septic patients are more sensitive to the hypotensive effects of drugs (25). Septic shock can also create a critical illness-related corticosteroid insufficiency, exacerbating a patient’s inability to generate appropriate systemic vascular resistance (26). While merely being hypotensive was not significantly associated with survival, the requirement for vasopressor therapy may be indicative of more severe disease. Patients in our study which received vasopressors had significantly higher preoperative lactate and higher rates of mortality. Prior veterinary studies have shown similar results, with survivors receiving fewer vasopressors, and those animals with septic shock having worse outcomes (5,10,27). In addition, Grimes et al (20) found that intraoperative hypotension was a risk factor for post-operative septic peritonitis and death in patients undergoing GI surgery. Also, human patients with septic shock are at higher risk of mortality than those with just severe sepsis (28).

Despite multiple methods of data analysis, we found no correlation between time from admission until surgery and survival, nor any correlation between duration of symptoms and survival. One possible explanation is that we did not have a large enough sample size to detect a difference, i.e., type II error. The small sample size could also potentially conceal significant disease-specific associations between surgical delay and outcome. Intra-operative euthanasia may have confounded the results in the case of patients which had a short TTI, but in which the owners were not expecting a diagnosis of septic peritonitis, and elected euthanasia based upon the change of projected cost or prognosis, or perceived post-operative quality of life in the mind of the owner. However we purposefully included patients euthanized during surgery because the decision to euthanize was often based on the surgeon’s recommendation regarding either unresectable masses or perceived post-operative quality of life. Furthermore, even with patients which were euthanized excluded from the analysis, there was still no significant difference in survival based upon TTI.

Controlled delay of surgery at the discretion of the primary surgeon for pre-operative stabilization/optimization may also have weakened the expected association between TTI and survival. Conversely, an overzealous urge to take patients to surgery prematurely could have contributed to the outcome.

Previous veterinary studies of septic peritonitis have shown a lack of association between outcome and duration of symptoms (3) or between outcome and time from admission until surgery (10). In addition, a study investigating the impact of early antibiotic administration could not demonstrate a significant difference in survival in patients with early administration of antibiotics (9).

While it seems intuitive that early surgical intervention should be paramount in these cases, the concept of rapid surgical source control is based mainly in dogma, as there is a lack of evidence regarding the ideal time for surgical intervention in septic peritonitis (28). In humans, many studies have shown that expeditious implementation of resuscitation bundles and early antimicrobial prophylaxis improve outcomes of patients with sepsis (29), but few studies have shown marked improvement in outcome with early surgical intervention (28). Furthermore, the majority of human studies on TTI examine necrotizing soft tissue infections and pancreatitis, while studies examining TTI in cases of abdominal perforation and subsequent sepsis are rare. Two relatively small studies (14,30) showed increased mortality in patients in whom surgery was delayed by 24 and 48 h, and a much larger retrospective cohort study found that in patients with a perforated gastric ulcer, there was an increase in the 30-day mortality of 2.4% per hour of surgical delay (31). However, that latter group recently published results of a slightly larger retrospective study of more recent cases that showed no association between surgical delay and outcome (32).

Facilitating shorter times from hospital entry until source control is critical, but rapid transfer to the operating room is not likely to succeed independently of proper resuscitation of septic patients, with end points of goal directed therapy that are targeted before surgical intervention. A semi-prospective study from Moore et al (33) found that when compared against national database of patients, a hospital with a dedicated “acute care surgery” service had improved survival rates, attributing this result to compliance with guidelines from the Surviving Sepsis campaign mandating early acquisition of cultures and baseline laboratory analyses, prompt goal directed fluid resuscitation, administration of antibiotics within 1 h of diagnosis, and rapid surgical source control (< 6 h).

A recent prospective study in humans with gastrointestinal perforation found that when early goal-directed therapy was combined with early surgical source control, time until surgery was a critical determinant of survival. There was a 0% survival rate in those patients for whom surgery occurred more than 6 h after admission. The authors concluded that humans with gastrointestinal perforation should be operated on within 1 to 2 h of diagnosis, regardless of a successful response to resuscitation efforts (34).

Study limitations include its retrospective nature and a low sample size. The data include cases managed by 7 surgeons, and this lack of standardized treatment protocols undoubtedly impacts outcome. Owners potentially contributed to the high rate of non-survival due to an inability or unwillingness to pursue expensive, intensive hospitalization in the face of a guarded prognosis. That confounding effect was minimized due to inclusion of only patients that were taken to surgery; however, there were some cases in which septic peritonitis was not discovered until surgery, requiring owners to rethink their decision to continue.

Mortality rates regarding septic peritonitis in this and other studies should be interpreted with caution because euthanasia was not distinguished from natural death, and intraoperative euthanasia or death was not distinguished from post-operative euthanasia or death. The decision to euthanize may have been made before definitive source control was attempted due to the perceived negative prognosis or anticipated cost and quality of life after extensive bowel resection. Prospective studies in which the reason for euthanasia is recorded would be helpful in order to better define the true prognosis with this condition. However, the survival rate in our study was 60%, which is similar to other reports with a comparable population.

In conclusion, we found that time from admission to surgical intervention in patients with septic peritonitis did not affect outcome and that several pre- and post-operative variables (age, PCV, ALP, bilirubin, lactate, albumin) appeared to have value in predicting outcome. These variables may be useful in future development of a prediction model or scoring system for patients with septic peritonitis and may help clinicians inform owners of prognosis prior to surgical intervention. To further investigate the importance of rapid source control, future prospective studies should focus not simply on expediting transfer to the operating room, but on better adherence to the Acute Care Surgery Model, in which patients with a diagnosis of septic peritonitis are treated in accordance with the evidence-based guidelines of the Surviving Sepsis Campaign. Computer-based decision-making protocols may help standardize and promote compliance (9,35). CVJ