A comparison of statistical and connectionist models for the prediction of chronicity in a surgical intensive care unit.

OBJECTIVE: To compare statistical and connectionist models for the prediction of chronicity which is influenced by patient disease and external factors. DESIGN: Retrospective development of predictive criteria and subsequent prospective testing of the same predictive criteria, using multiple logistic regression and three architecturally distinct neural networks; revision of predictive criteria. SETTING: Surgical intensive care unit (ICU) equipped with a clinical information system in a +/- 1000-bed university hospital. PATIENTS: Four hundred ninety-one patients with ICU length of stay 3 days who survived at least an additional 4 days. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Chronicity was defined as a length of stay > 7 days. Neural networks predicted chronicity more reliably than the statistical model regardless of the former's architecture. However, the neural networks' ability to predict this chronicity degraded over time. CONCLUSIONS: Connectionist models may contribute to the prediction of clinical trajectory, including outcome and resource utilization, in surgical ICUs.

Polyethylene glycol-conjugated superoxide dismutase fails to augment brain superoxide dismutase activity in piglets.

We studied the effect of intravenously administered polyethylene glycol-conjugated superoxide dismutase (8,000 units/kg) on brain superoxide dismutase activity in 44 1-2-week-old piglets in the absence and presence of global cerebral ischemia and reperfusion. Four groups (n = 6 each) of piglets not exposed to ischemia were studied. Enzyme administration increased plasma superoxide dismutase activity from less than 5 to 142 +/- 8 units/ml (mean +/- SEM) without increasing brain activity (e.g., activities in the caudate were 7.9 +/- 0.5 and 8.1 +/- 0.4 units/mg protein) for up to 2 hours following administration. Four additional groups (n = 5 each) of piglets were given either enzyme or polyethylene glycol 5 minutes prior to 10 minutes of global cerebral ischemia induced by aortic cross-clamping followed by either 5 or 45 minutes of reperfusion. Enzyme administration increased plasma superoxide dismutase activity from less than 5 to 144 +/- 5 units/ml but failed to increase brain activity even after 45 minutes of reperfusion (e.g., activities in the caudate were 8.5 +/- 0.3 and 8.6 +/- 0.6 units/mg protein). We conclude that intravenous polyethylene glycol-conjugated superoxide dismutase does not increase superoxide dismutase activity in the brain despite global ischemia and reperfusion.

Effect of adrenal hypotension on elicited secretory activity in anesthetized dogs.

The objective of this study was to determine whether three- to fourfold increases in adrenal medullary blood flow (MQ), observed during catecholamine secretion induced by splanchnic nerve stimulation (NS), are necessary for a maximal secretory response. Eight pentobarbital-anesthetized adult male mongrel dogs (25-35 kg) were subjected to two episodes of NS-induced catecholamine secretion. To limit increases in MQ, one NS (20 Hz, 8 V, 1.5 ms) was performed under control conditions, the other was after reduction of adrenal perfusion pressure (PP). PP was 124.8 +/- 6.0 at control and was reduced to 23.0 +/- 1.0 mmHg by inflation of an aortic occluder located 3 in. above the adrenals. NS under control conditions increased ipsilateral MQ (measured using radiolabeled microspheres) from 181 +/- 39 to 1,336 +/- 199 ml.min-1.100 g-1 and adrenal epinephrine secretion from 184 +/- 128 to 11,445 +/- 5,216 ng.min-1.g medulla-1. Contralateral MQ and cortical blood flow were unaffected by NS. Reduction of PP decreased MQ to 32 +/- 6 ml.min-1.100 g-1 and adrenal cortical blood flow from 251 +/- 23 to 25 +/- 6 ml.min-1.100 g-1.PP reduction had no effect on either corticosteroid secretion or unstimulated epinephrine secretion. NS at reduced PP increased ipsilateral MQ to 146 +/- 27 ml.min-1.100 g-1 and epinephrine secretion to 13,935 +/- 5,175 ng.min-1.g medulla-1. These data indicate that epinephrine secretion is not altered when MQ is limited by reduction of PP and suggest that, at normal PP, increases in MQ during catecholamine secretion are not necessary for a maximal secretory response.(ABSTRACT TRUNCATED AT 250 WORDS)

Aminophylline does not attenuate histamine-induced airway constriction during halothane anesthesia.

The effects of aminophylline on the release of endogenous catecholamines and on airway reactivity to aerosol histamine challenge were evaluated during halothane and thiopental/fentanyl anesthesia in basenji-greyhound dogs. Responses to histamine aerosol challenge (0.01, 0.03, 0.1, 0.3, 1.0, and 3.0 mg/ml) were measured during six conditions: 1) thiopental/fentanyl anesthesia (control), 2) thiopental/fentanyl with aminophylline infusion, 3) halothane anesthesia (1.5 MAC), 4) halothane anesthesia with aminophylline infusion, 5) thiopental/fentanyl anesthesia after pretreatment with iv propranolol, and 6) thiopental/fentanyl anesthesia with aminophylline infusion after pretreatment with iv propranolol. Prior to aerosol challenge baseline pulmonary resistance (RL) did not differ in the six groups. Aminophylline significantly attenuated the pulmonary response to histamine and increased catecholamine concentrations during thiopental/fentanyl anesthesia. Although halothane itself significantly attenuated the pulmonary response to histamine, the administration of aminophylline during halothane anesthesia produced no additional protective effect and no increases in catecholamines were noted. Moreover, no protective effect was seen after aminophylline administration during thiopental/fentanyl anesthesia in the same dogs pretreated with propranolol. These data suggest that the protective effect of aminophylline on histamine reactivity results from release of endogenous catecholamines and that the use of aminophylline during halothane anesthesia, which blocks this release, is not warranted.

Ventricle-brain ratio, computed tomographic density, and brain area in 50 schizophrenics.

A sample of 50 DSM-III-diagnosed schizophrenics (mean age, 34 years) intentionally biased to contain a relatively high proportion of persistently unemployed persons was compared with a sample of 87 normal volunteers on three computed tomographic measures. These were lateral ventricle-brain ratio, regional brain computed tomographic density values, and brain slice area. Images were made with the same computed tomographic scanner and identical scan parameters. Computed tomographic data were assessed blindly using a computer-linked image array processor and electronic planimeter. Ventricle-brain ratios were significantly higher in schizophrenics, with 28% of the patient sample exceeding 2 SDs of the control mean. Brain area measures were not associated with an enlarged ventricle-brain ratio. Contrary to our prediction, ventricular enlargement was unassociated with most negative symptom ratings, but was correlated with the absence of positive symptoms. A history of abnormal delivery and the presence of left-handedness were significant predictors of an enlarged ventricle-brain ratio on multiple regression analysis. Schizophrenics had a significantly smaller brain slice area compared with normal controls, a finding not attributable to height differences between groups. Brain slice area was inversely correlated with computed tomographic brain density across all subjects. After correction of computed tomographic density values for area using a linear regression model, no significant regional density differences were detectable between normal controls and schizophrenics. Within normal controls there was a significant relationship between social class and brain slice area, but not ventricle-brain ratio.

Adrenergic receptors of adrenal medullary vasculature.

The present study evaluates possible effects of adrenal catecholamines, released by splanchnic nerve stimulation, on adrenal medullary blood flow (MQ) and adrenal catecholamine secretion (CS). Twelve pentobarbital-anesthetized mongrel dogs were subjected to three identical splanchnic nerve stimulations (5 V, 20 Hz, for 3 min) at 30-min intervals, and MQ (radiolabeled microsphere technique) and CS (high-performance liquid chromatography) were measured before and during each nerve stimulation. Animals were assigned to one of three groups and administered either saline, pindolol (1 and 4 mg/kg), or prazosin (1 and 4 mg/kg) before the second and third nerve stimulation, respectively. In the saline control group, each nerve stimulation resulted in similar increases in MQ and CS. Pindolol attenuated nerve stimulation-induced increases in MQ and CS by 50%, but had no effect on medullary catecholamine concentration. Prazosin augmented nerve stimulation-induced MQ, CS, and medullary catecholamine concentration by 35%. These data suggest that adrenal adrenergic receptors modulate elicited CS and mediate changes in adrenal medullary vascular tone.

Evaluation of sepsis in a critically ill surgical population.

We report a new clinical rating system which assesses septic patients' ongoing disease course and its severity. Our system incorporates the Therapeutic Intervention Scoring System (TISS) and Acute Physiology and Chronic Health Evaluation to measure discrete organ system abnormalities, plus a multiple system organ failure scale to quantify the number of abnormal organ systems. The resulting score, which reflects the severity of multiple organ dysfunction and grades responsiveness to therapy, was validated against the actual disease course. Retrospective and prospective profiles of individual surgical ICU patients demonstrated that this tracking method was a more effective indicator of severity of sepsis and more sensitive to the day-to-day changes in clinical status than either the TISS or APACHE II components alone. We also demonstrate that a graphic illustration of daily system scores yields clinically useful information relevant to the patients' septic course.

Differential and asymmetrical behavioral effects of electrolytic or 6-hydroxydopamine lesions in the nucleus accumbens.

Sprague-Dawley rats received either unilateral or bilateral electrolytic or 6-hydroxydopamine (6-OHDA) lesions in the nucleus accumbens (n. acc.) and daily spontaneous activity was measured in home cage running wheels for 30 days postoperatively. Bilateral electrolytic lesions increased activity to 230% of preoperative baseline. Right electrolytic lesions produced a 50% increase in activity while left lesion rats were not significantly different from sham-operated controls. Conversely, 6-OHDA lesions depressed activity. Bilateral 6-OHDA lesions had the greatest depressant effect upon activity while unilateral left or right n. acc. treatments produced relatively less hypoactivity. Results support an asymmetrically organized serial inhibition model of spontaneous locomotor control.

Cold water stress abolishes hyperactivity produced by cortical suction lesions without altering noradrenergic depletion.

The effect of forced cold water (15 degrees C) swimming on catecholamine depletions and behavioral hyperactivity produced by focal right hemispheric cortical suction lesions was assessed. Through the 30-day postoperative period, lesion and sham-operated control animals were placed in cold water for 5 min either four times at weekly intervals, once during the first postoperative week, or not at all. The effect of this cold water swim test was different for lesion and control animals. Stress at weekly intervals produced hyperactivity in control animals, but weekly stress in animals with focal suction lesions blocked the development of their expected hyperactivity. A single stress experience, 1 week postoperative, was also sufficient to block the development of hyperactivity in the lesion group. Stress, although affecting behavior, did not alter the pattern of norepinephrine depletions in animals receiving right hemispheric cortical suction lesions.

Social functioning assessment in stroke patients.

The influence of depression, intellectual function, and physical impairment on patient reports of social functioning was assessed by comparing responses obtained from 30 stroke patients with those obtained from an outside informant who knew the patient well. There was relatively good agreement between patient and other for both the overall score on the Social Functioning Examination (SFE) and individual items related to specific aspects of social functioning. Differences between scores obtained from a patient and outside informant were not significantly related to either depression, moderate degrees of intellectual impairment, relative closeness of patient and informant or to degree of physical impairment. These data suggest that valid SF assessments can be made either by a stroke patient who is capable of being interviewed or by a familiar outside informant. In addition, the prognostic utility of the SFE was examined by interviewing 50 patients during the acute stroke period and following them over six months. Social functioning in-hospital scores were not found significantly related to 6-month scores for either depression, intellectual impairment, or physical impairment, but depression and in-hospital impairment significantly predicted social functioning at 6-month follow-up. These data suggest that the most impaired stroke patients are the most likely to undergo social deterioration during the post-stroke period and may require the greatest amount of social intervention. Whether treatment of these variables significantly affects 6-month outcome remains an intriguing question for further study.

Asymmetrical effect of unilateral cortical lesions and amphetamine on DRL-20: a time-loss analysis.

Male Sprague Dawley rats received unilateral 1.5 mm diameter focal suction lesions of either the left or right orbito-frontal cortex and were tested for response to 6 doses of amphetamine sulfate ranging from 0.5 to 1.5 mg/kg on a DRL-20 schedule of reinforcement. Right hemispheric lesion animals obtained a greater number of control reinforcements and were more sensitive to amphetamine's disruptive effects, showing a greater dose-related decrease in water rewards obtained than left lesion animals. An analysis method is introduced which combines the interactive effects of premature responses and their IRT value in a way that relates directly to reinforcement attainment. Calculated total session time made unavailable for reinforcement due to premature responding, correlated negatively (r = -.942) with the number of reinforcements obtained.

Marked behavioral and biochemical sensitivity to lesion size in the posterior cortex of the rat.

Unilateral focal cortical suction lesions of varying diameter from 0.7 to 2.0 mm were made in the right posterior lateral cortex of rats. Only the 1.0 mm lesion size resulted in spontaneous postoperative hyperactivity. This increased activity was accompanied by significant elevations in substantia nigra and caudate nucleus dopamine concentrations. Other lesion sizes did not affect behavior or brain biochemical concentrations in this fashion. These results suggest a functional organization in this area of posterior cortex which affects both neurochemistry and behavior.

Automated measurement of rearing behavior in adult and neonatal rats.

Three variables constituting an index of rearing behavior from an animal-activity monitoring system were compared with a standard visual rating at various ages of neonatal and adult rats. All three automated measures correlated significantly with measures rated by visual observation, although only the "number of vertical episodes" was similar in magnitude to the actual number of observed rearings. The height of the vertical sensors was critical in detecting rearings, although reliable results were obtained over a relatively wide range of heights. Analyses in neonatal rats demonstrated developmental progression in rearing behavior, with an onset around Day 15 of age. This automated technique was found to be useful and eliminated confounding variables, such as experimenter bias, commonly associated with visual observation techniques.

Asymmetrical effects of unilateral cortical suction lesions: behavioral characterization.

Behavioral effects of unilateral frontocortical suction lesions in the right and left hemisphere were characterized by a computerized activity monitoring system. Lesions of the right frontal cortex produced postoperative hyperactivity consisting of increases in the total distance traveled, average distance per movement, and average speed of movement. Vertical, rotational, and stereotypic measures were not changed. Animals receiving left hemisphere lesions did not differ from sham-operated controls on any behavioral variable. The lateralized hyperactivity resulting from lesions of the right frontal cortex was confined to increased length and speed of horizontal movement and did not appear to represent a generalized increase in all components of spontaneous activity.

Asymmetrical locomotor response to unilateral cortical injections of DSP-4.

The role of norepinephrine depletion in the lateralized production of spontaneous hyperactivity was assessed by unilateral fronto-cortical injections of either 10 or 20 micrograms of the noradrenergic neurotoxin, DSP-4 (N-2-chlorethyl-N-ethyl-bromobenzylamine hydrochloride). Ten micrograms of DSP-4 produced significant hyperkinesis only when injected into the right hemisphere. A 20 micrograms dose produced hyperactivity when injected into either hemisphere. DSP-4 injections resulted in significant NE and ipsilateral and contralateral 5HT depletions in the frontal cortex. The 20 micrograms right hemispheric injection significantly increased both DOPAC and DA levels in the contralateral caudate in a manner unrelated to behavior. These findings, in conjunction with results from previous neurotoxin studies, support a hypothesis of greater relative sensitivity to injury of right hemispheric NE terminal fields compared with injury to left hemispheric fields as demonstrated by spontaneous hyperactivity.

Cortical undercuts in the rat produce asymmetrical behavioral response without altering catecholamine concentrations.

Baseline spontaneous activity of male Sprague-Dawley rats was established in home running wheel cages 3 weeks prior to cortical undercutting. Rats received unilateral 2-mm-diameter circular lesions undercutting layer VI of either the left or right frontal cortex. The right hemisphere lesion group showed a 63% increase in spontaneous activity and rats with left hemisphere lesions were no more active than sham-operated controls. No significant intergroup differences in norepinephrine concentrations were noted in the frontal or posterior cortex or locus ceruleus either ipsilateral or contralateral to the lesion site. In addition, there were no significant group differences in dopamine content of the caudate nucleus. Results suggest that the lateralized neural substrates controlling locomotor asymmetry are probably postsynaptic to cortical norepinephrine fibers and may involve glutamatergic corticostriatal projections.

Mood disorders in stroke patients. Importance of location of lesion.

In a selected group of right-handed patients with single stroke lesions of either the right (n = 14) or left (n = 22) hemisphere and no predisposing factors for psychiatric disorder, we found that the severity of depression was significantly increased in patients with left anterior lesions as opposed to any other lesion location. In addition, the severity of depression correlated significantly with proximity of the lesion on CT scan to the frontal pole in the left anterior group. The right hemisphere lesion group showed the reverse trend: patients with right posterior lesions were more depressed than patients with right anterior lesions, who were unduly cheerful and apathetic. These findings suggest that intrahemispheric lesion location is in some way related to mood disorder in stroke patients and that there is a graded effect of lesion location on severity of mood change. The neuroanatomy of the biogenic amine-containing pathways in the cerebral cortex might explain this graded effect and provide a neurochemical basis for the mood change.

Effect of anterior-posterior lesion location on the asymmetrical behavioral and biochemical response to cortical suction ablations in the rat.

Small cortical suction ablations were produced at one of several stereotaxically located sites along the anterior-posterior axis of the right or left hemisphere in the rat. Analysis of variance showed a highly significant effect of lesion location in the right hemisphere on locomotor activity. The most anterior lesions produced both the most hyperactivity and also the greatest reductions in the concentrations of norepinephrine in the ipsilateral and contralateral cortex and locus coeruleus. These results suggest that the effect of cortical lesions on spontaneous activity may be graded across the right hemisphere and that the anatomy of certain neurotransmitter pathways in the cortex may help to explain both the biochemical and behavioral findings.

Asymmetrical effects of cortical island lesions in the rat.

Baseline spontaneous activity of male Sprague-Dawley rats was established in home running wheel cages 3 weeks prior to receiving cortical circumscription lesions. Circumscriptions 2 mm in diameter and 2 mm deep were produced by means of a rotating microknife and were centered approximately 9.0 mm anterior to ear bar zero. Circular knife cuts of the right, but not the homologous left cortex produced long-lasting depletions of ipsilateral and contralateral fronto- and posterio-cortical NE. Right hemisphere lesions produced a 35% increase in activity while left hemispheric lesions uniformly decreased spontaneous home cage running wheel activity by 35%. These results are consistent with our hypothesis that injury to transcortical pathways (perhaps noradrenergic fibers) are sufficient to produce lateralized spontaneous hyperactivity. The possible existence of competing excitatory and inhibitory mechanisms in the left frontal cerebral cortex is discussed.