Anesthesiology

With appropriate algorithms, computers can learn to detect patterns and associations in large data sets. The authors’ goal was to apply machine learning to arterial pressure waveforms and create an algorithm to predict hypotension. The algorithm detects early alteration in waveforms that can herald the weakening of cardiovascular compensatory mechanisms affecting preload, afterload, and contractility.

The algorithm was developed with two different data sources: (1) a retrospective cohort, used for training, consisting of 1,334 patients’ records with 545,959 min of arterial waveform recording and 25,461 episodes of hypotension; and (2) a prospective, local hospital cohort used for external validation, consisting of 204 patients’ records with 33,236 min of arterial waveform recording and 1,923 episodes of hypotension. The algorithm relates a large set of features calculated from the high-fidelity arterial pressure waveform to the prediction of an upcoming hypotensive event (mean arterial pressure < 65 mmHg). Receiver-operating characteristic curve analysis evaluated the algorithm’s success in predicting hypotension, defined as mean arterial pressure less than 65 mmHg.

Using 3,022 individual features per cardiac cycle, the algorithm predicted arterial hypotension with a sensitivity and specificity of 88% (85 to 90%) and 87% (85 to 90%) 15 min before a hypotensive event (area under the curve, 0.95 [0.94 to 0.95]); 89% (87 to 91%) and 90% (87 to 92%) 10 min before (area under the curve, 0.95 [0.95 to 0.96]); 92% (90 to 94%) and 92% (90 to 94%) 5 min before (area under the curve, 0.97 [0.97 to 0.98]).

The results demonstrate that a machine-learning algorithm can be trained, with large data sets of high-fidelity arterial waveforms, to predict hypotension in surgical patients’ records.

Nhiệt độ da nên được giữ ổn định khi xác định ngưỡng phản ứng vì cả nhiệt độ da và nhiệt độ lõi đều đóng góp vào việc kiểm soát điều hòa nhiệt. Tuy nhiên, trong thực tế, khó khăn trong việc đánh giá cả ngưỡng nóng và lạnh trong khi vẫn giữ nhiệt độ da không thay đổi. Một nghiên cứu gần đây cho thấy ngưỡng co mạch và run lắc là một hàm tuyến tính của nhiệt độ da và nhiệt độ lõi, với nhiệt độ da đóng góp 20 +/- 6% và 19 +/- 8%, tương ứng. (Nhiệt độ da đã được biết đến từ lâu là đóng góp khoảng 10% vào việc kiểm soát việc toát mồ hôi). Sử dụng các mối quan hệ này, chúng tôi đã tiến hành điều chỉnh thực nghiệm cả nhiệt độ da và nhiệt độ lõi, sau đó bù đắp cho những thay đổi trong nhiệt độ da, và cuối cùng báo cáo kết quả theo các ngưỡng nhiệt độ lõi được tính toán tại một nhiệt độ da quy định duy nhất.

Năm tình nguyện viên đã được nghiên cứu trong 4 ngày: (1) nhóm đối chứng; (2) nồng độ propofol mục tiêu trong máu là 2 microgam/ml; (3) nồng độ mục tiêu là 4 microgam/ml; và (4) nồng độ mục tiêu là 8 microgam/ml. Vào mỗi ngày, chúng tôi đã tăng nhiệt độ da và nhiệt độ lõi đủ để gây ra việc toát mồ hôi. Sau đó, nhiệt độ da và nhiệt độ lõi được giảm xuống để kích thích co mạch ngoại vi và run lắc. Chúng tôi đã bù đắp toán học cho những thay đổi trong nhiệt độ da bằng cách sử dụng các đóng góp tuyến tính đã được thiết lập cho việc kiểm soát việc toát mồ hôi (10%) và cho co mạch và run lắc (20%). Từ những ngưỡng nhiệt độ lõi được tính toán này (tại một nhiệt độ da quy định là 35.7 độ C), các đường cong phản ứng nồng độ propofol cho ngưỡng toát mồ hôi, co mạch và run lắc đã được phân tích bằng cách sử dụng hồi quy tuyến tính. Chúng tôi đã xác thực phương pháp mới này bằng cách so sánh các tác động phụ thuộc vào nồng độ của propofol với các kết quả đã thu được trước đó với một mô hình đã được thiết lập.

This study was conducted to investigate the feasibility of using ultrasound as an image tool to locate the sacral hiatus accurately for caudal epidural injections.

Between August 2002 and July 2003, 70 patients (39 male and 31 female patients) with low back pain and sciatica were studied. Soft tissue ultrasonography was performed to locate the sacral hiatus. A 21-gauge caudal epidural needle was inserted and guided by ultrasound to the sacral hiatus and into the caudal epidural space. Proper needle placement was confirmed by fluoroscopy.

In all the recruited patients, the sacral hiatus was located accurately by ultrasound, and the caudal epidural needle was guided successfully to the sacral hiatus and into the caudal epidural space. There was 100% accuracy in caudal epidural needle placement into the caudal epidural space under ultrasound guidance as confirmed by contrast dye fluoroscopy.

Ultrasound is radiation free, is easy to use, and can provide real-time images in guiding the caudal epidural needle into the caudal epidural space. Ultrasound may therefore be used as an adjuvant tool in caudal needle placement.

Recent studies show that intraoperative mechanical ventilation using low tidal volumes (VT) can prevent postoperative pulmonary complications (PPCs). The aim of this individual patient data meta-analysis is to evaluate the individual associations between VT size and positive end–expiratory pressure (PEEP) level and occurrence of PPC.

Randomized controlled trials comparing protective ventilation (low VT with or without high levels of PEEP) and conventional ventilation (high VT with low PEEP) in patients undergoing general surgery. The primary outcome was development of PPC. Predefined prognostic factors were tested using multivariate logistic regression.

Fifteen randomized controlled trials were included (2,127 patients). There were 97 cases of PPC in 1,118 patients (8.7%) assigned to protective ventilation and 148 cases in 1,009 patients (14.7%) assigned to conventional ventilation (adjusted relative risk, 0.64; 95% CI, 0.46 to 0.88; P < 0.01). There were 85 cases of PPC in 957 patients (8.9%) assigned to ventilation with low VT and high PEEP levels and 63 cases in 525 patients (12%) assigned to ventilation with low VT and low PEEP levels (adjusted relative risk, 0.93; 95% CI, 0.64 to 1.37; P = 0.72). A dose–response relationship was found between the appearance of PPC and VT size (R2 = 0.39) but not between the appearance of PPC and PEEP level (R2 = 0.08).

These data support the beneficial effects of ventilation with use of low VT in patients undergoing surgery. Further trials are necessary to define the role of intraoperative higher PEEP to prevent PPC during nonopen abdominal surgery.

Cannulation of the internal jugular vein (IJV) is associated with a 95% success rate when external landmarks are used. Anatomic variability has been implicated as the cause for difficulty in cannulation without ultrasound. In contrast to an IJV located lateral to the carotid artery (CA), an IJV overlying the CA may result in CA puncture. The authors' purpose in this study was to examine, using ultrasound, the anatomic relation of the IJV and CA as viewed from the perspective of a cannulating needle.

Ultrasound imaging was used in 1,136 patients to examine the relation between the IJV and CA. A 7.5-MHz transducer was placed in the direction of a cannulating needle on the right neck at the apex of the angle formed by the division of the sternocleidomastoid muscle. A Polaroid photograph of the image was later scored by three blinded investigators according to the percentage of the CA overlaid by the IJV (0 to 4).

Of the 1,136 Polaroid photographs of the ultrasound images, 1,009 were suitable for scoring. Fifty-four percent of all patients received a score of 4, which indicated that the IJV overlies more than 75% of the CA in an imaging plane positioned in the direction of a cannulating needle. Patients older than 60 yr were more likely to have this anatomy than patients younger than 60 yr (P < 0.05). None of the other patient characteristics recorded were predictive of this anatomic relation.

In a majority of patients, the IJV is not lateral to the CA in an ultrasound imaging plane positioned in the direction of a cannulating needle. Instead, the IJV overlies the CA in 54% of patients overall, predisposing these patients to CA puncture if the cannulating needle traverses the IJV.

Intrathecal injection of clonidine and dexmedetomidine produce behavioral analgesia by an alpha 2-adrenergic mechanism. Functional and anatomic studies suggest that this analgesia is mediated by cholinergic activation. This hypothesis was directly tested by measuring extracellular acetylcholine concentrations in spinal cord interstitial fluid by means of microdialysis after intrathecal injection of these alpha 2-adrenergic agonists in sheep.

Twelve sheep with chronically implanted thoracic intrathecal catheters were anesthetized with halothane. Multiple 200-micron-diameter dialysis fibers were inserted surgically at a mid-thoracic level through the dorsal horn and perfused with artificial cerebrospinal fluid. After baseline sampling, either clonidine (100 micrograms), dexmedetomidine (100 micrograms), or saline were injected intrathecally. Microdialysis samples were analyzed by high-pressure liquid chromatography for acetylcholine and norepinephrine.

Both alpha 2-adrenergic agonists increased acetylcholine in microdialysate, whereas intrathecal saline had no effect. Analysis of the raw data showed that all groups differed significantly, with greater levels of acetylcholine following administration of dexmedetomidine than clonidine or saline. Unexpectedly, intrathecal clonidine also increased microdialysate norepinephrine levels.

These data are consistent with previous experiments measuring acetylcholine concentrations in cerebrospinal fluid and support analgesia from alpha 2-adrenergic agonists mediated in part by cholinergic activation. In addition, the increase in norepinephrine concentrations after intrathecal administration of clonidine suggest stimulation of norepinephrine release by this agent.

Opioids produce analgesia by direct effects as well as by activating neural pathways that release nonopioid transmitters. This study tested whether systematically administered opioids activate descending spinal noradrenergic and cholinergic pathways.

The effect of intravenous morphine on cerebrospinal fluid and dorsal horn microdialysate concentrations of norepinephrine and acetylcholine was examined in 20 sheep. Animals received either intravenous morphine or fentanyl alone, or morphine plus intravenous naloxone or intrathecal idazoxan.

Intravenous morphine (0, 0.5, 1 mg/kg, intravenous) produced dose-dependent increases in cerebrospinal fluid norepinephrine and acetylcholine, but not epinephrine or dopamine. Morphine's effect was blocked by intravenous naloxone and by intrathecal idazoxan. In microdialysis experiments, intravenous morphine increased the concentration of norepinephrine and acetylcholine, but not epinephrine or dopamine, in the dorsal horn. In contrast, intravenous morphine exerted no effect on any of these monoamines in the ventral horn. Intravenous naloxone and cervical cord transection each blocked morphine's effect on dorsal horn norepinephrine.

These results support functional studies that indicate that systematically administered opioids cause spinal norepinephrine and acetylcholine release by a naloxone-sensitive mechanism. Idazoxan blockade of morphine's effects on cerebrospinal fluid norepinephrine was unexpected, and suggests that both norepinephrine and acetylcholine release in the spinal cord may be regulated by alpha 2-adrenoceptors. Microdialysis experiments suggest increased norepinephrine and acetylcholine levels in cerebrospinal fluid resulted from intravenous morphine-induced activation of bulbospinal pathways.

Nitric oxide synthase is located in the spinal cord dorsal horn and intermediolateral cell column, where it may modulate sensory and sympathetic neuronal activity. However, the biochemical characteristics of this enzyme have not been examined in these different areas in the spinal cord. Although alpha(2)-adrenergic agonists, muscarinic agonists, and nitric oxide may interact in the spinal cord to produce antinociception, these interactions have not been characterized.

Sheep spinal cord tissue was homogenized ad centrifuged at high sped to separate soluble and membrane-bound fractions. Nitric oxide synthase activity was determined by conversion of [(14)C]-L-arginine to [(14)C]-L-citrulline and its kinetic characteristics, dependency on cofactors, and sensitivity to inhibitors determined. Sheep spinal cord was stained for nicotinamide adenine dinucleotide phosphate diaphorase as a marker for nitric oxide synthase. Antinociception to a mechanical stimulus from intrathecal clonidine alone and with neostigmine was determined and the effects of L-arginine and n-methyl-L-arginine were determined.

More than 85% of nitric oxide synthase activity was present in the soluble form and its kinetic, cofactor, and antagonist properties were similar to those of the neuronal isoform of nitric oxide synthase. Biochemical and histochemical studies localized nitric oxide synthase to the superficial dorsal horn and the intermediolateral cell column. Clonidine antinociception was enhanced by L-arginine and neostigmine, but not by D-arginine. Neostigmine's enhancement of clonidine antinociception was blocked by n-methyl-L-arginine.

These results confirm those of previous studies demonstrating localization of nitric oxide synthase to superficial dorsal horn and intermediolateral cell column of mammalian spinal cord, and suggesting its identity as the neuronal isoform. Spinal alpha(2)-adrenergic agonist antinociception may be partly dependent on cholinergic and nitric oxide mechanisms.

alpha 2-Adrenergic agonists are thought to produce analgesia, in part, by activating spinal acetylcholine release. The purpose of the current study was to examine the interaction between intrathecal neostigmine and epidural clonidine for analgesia and side effects in humans.

A total of 58 volunteers received an intrathecal injection of 5% dextrose in normal saline (D5NS) or neostigmine (50, 100, or 200 micrograms in D5NS), followed in 1 h by epidural saline or clonidine (computer-controlled infusion targeted to 50, 100, 200, or 400 ng/ml in cerebrospinal fluid) using an isobolographic design. Visual analog scale pain to a noxious cold stimulus, nausea, weakness, sedation, and other safety variables was measured before and at specified intervals after drug administration.

The first 21 volunteers randomized to receive intrathecal hyperbaric neostigmine rather than D5NS received the drug while in the sitting position, and had none-to-minimal analgesia 1 h later. The remaining volunteers received the drug while in the lateral position, and demonstrated dose-dependent analgesia in the foot 1 h later. Epidural clonidine also caused dose-dependent analgesia. The combination of neostigmine and clonidine resulted in an additive enhancement for analgesia, but no enhancement of each drug's side effects, and a reduction in clonidine-induced hypotension. Neostigmine injected into subjects in the lateral position diminished clonidine-induced reductions in blood pressure and plasma norepinephrine.

These results support enhancement of alpha 2-adrenergic analgesia by intrathecal neostigmine, but do not demonstrate synergy, as observed in animals. Lack of enhancement of side effects suggests this combination may be clinically useful.

In dogs, sheep, and rats, spinal neostigmine produces analgesia alone and enhances analgesia from alpha 2-adrenergic agonists. This study assesses side effects and analgesia from intrathecal neostigmine in healthy volunteers.

After institutional review board approval and informed consent, 28 healthy volunteers were studied. The first 14 volunteers received neostigmine (50-750 micrograms) through a #19.5 spinal needle followed by insertion of a spinal catheter. The remaining 14 volunteers received neostigmine through a #25 or #27 spinal needle without a catheter. Safety measurements included blood pressure, heart rate, oxyhemoglobin saturation, end-tidal carbon dioxide, neurologic evaluation, and computer tests of vigilance and memory. Analgesia in response to ice water immersion was measured.

Neostigmine (50 micrograms) through the #19.5 needle did not affect any measured variable. Neostigmine (150 micrograms) caused mild nausea, and 500-750 micrograms caused severe nausea and vomiting. Neostigmine (150-750 micrograms) produced subjective leg weakness, decreased deep tendon reflexes, and sedation. The 750-micrograms dose was associated with anxiety, increased blood pressure and heart rate, and decreased end-tidal carbon dioxide. Neostigmine (100-200 micrograms) in saline, injected through a #25 or #27 needle, caused protracted, severe nausea, and vomiting. This did not occur when dextrose was added to neostigmine. Neostigmine by either method of administration reduced visual analog pain scores to immersion of the foot in ice water.

The incidence and severity of these adverse events from intrathecal neostigmine appears to be affected by dose, method of administration, and baricity of solution. These effects in humans are consistent with studies in animals. Because no unexpected or dangerous side effects occurred, cautious examination of intrathecal neostigmine alone and in combination with other agents for analgesia is warranted.