To determine the effect of general anesthesia on the accommodative tone in children.
Cohort study in an academic practice.
In children under 12 years of age who were undergoing general anesthesia, cycloplegic refraction was measured using streak retinoscopy during an office visit. Within 6 months, streak retinoscopy without cycloplegia was performed under general anesthesia. The main outcome measure was the difference between retinoscopy under anesthesia and cycloplegic retinoscopy in children.
In 41 children with an average age of 3.7 years (range, 0.8 to 11 years) retinoscopy under anesthesia yielded significantly more myopic measurements than cycloplegic retinoscopy for the sphere and spherical equivalent ( P < 0.0001 for both) but was in good agreement with cycloplegic retinoscopy for cylinder power and axis. The average difference between retinoscopy under anesthesia and cycloplegic retinoscopy was −0.98 diopters (D) (95% limit of agreement, −3.08 D to +1.10 D) for the sphere, 0.08 D (95% limit of agreement, −0.67 D to +0.82 D) for the cylinder, and −0.94 D (95% limit of agreement, −3.01 D to +1.13 D) for the spherical equivalent. Retinoscopy under anesthesia was within 1 D of cycloplegic retinoscopy in 25 subjects (61%) for the sphere, in all subjects for the cylinder, and in 28 subjects (68.3%) for the spherical equivalent.
Although general anesthesia reduced the accommodative tone in most children, it was still significant in some as compared to the tone found in cycloplegic retinoscopy. If an accurate measurement is essential, cycloplegia is recommended when measuring refraction in children under general anesthesia.
To measure the refractive error in children, streak retinoscopy is used routinely. Cycloplegic eyedrops are usually applied to paralyze accommodation and allow for accurate measurements. Without cycloplegia, accommodation can lead to overestimation and overdiagnosis of myopia. In some children, the refractive error is measured during an examination under general anesthesia, but the effect of current general anesthesia protocols on accommodative tone in children is not well understood. Earlier studies reported general anesthesia to induce a myopic shift of as much as 5 diopters (D). Induced anisometropia and instability in the spherical component of the refractive error have also been reported. General anesthesia protocols used in earlier studies include the parenteral administration of atropine, which is a potent cycloplegic agent; in current general anesthesia protocols, atropine is rarely administered.
Several stages of general anesthesia have been defined, although some controversy exist regarding these definitions :
Stage 1: Awake, not anesthetized;
Stage 2: Partially anesthetized, unconscious but delirious, involuntary movements, dysconjugate gaze, great possibility of laryngospasm or vomiting, irregular breathing;
Stage 3: Surgical anesthesia, immobile, conjugate midline gaze, regular breathing;
Stage 4: Anesthetic overdose, impending cardiovascular collapse.
This staging system is commonly used with newer anesthetic agents (sevoflurane, isofurane and desflurane). For ophthalmic surgery or examinations under general anesthesia, stage 3 is achieved and maintained throughout the procedure.
If general anesthesia induces complete relaxation of the accommodative tone, the use of cycloplegic eyedrops would be unnecessary when measuring the refractive error under general anesthesia. In order to investigate the effect of current general anesthesia on the accommodative tone in children, we compared the retinoscopy performed in the office after application of cycloplegic eyedrops (cycloplegic retinoscopy) with the retinoscopy performed under general anesthesia without prior administration of cycloplegic eyedrops (retinoscopy under anesthesia).
The protocol for this cohort study adhered to the tenets of the Declaration of Helsinki. It was approved by the Yale University School of Medicine institutional review board and is in accordance with all Health Insurance Portability and Accountability Act (HIPAA) regulations.
Consent to participate in this study was obtained from the subject’s parents or guardians; children 7 years and older also assented to participation. Over a 27-month period, children ranging from 6 months to 11 years of age who were scheduled to have general anesthesia for ophthalmic procedures were invited to participate in the study. The upper limit of 11 years was chosen because examinations under general anesthesia are most commonly performed in the younger age group and because accommodation is known to decline with age. Exclusion criteria for this study were cycloplegic retinoscopy more than 6 months prior to general anesthesia; poor cooperation during cycloplegic retinoscopy in the office; disorders of accommodation; prior or current atropine treatment of the study eye (eg, for amblyopia) or any other cycloplegic eyedrop immediately preoperatively; and underlying neurologic abnormalities. If retinoscopy under anesthesia could not be completed within five minutes, it was aborted to avoid extending general anesthesia for the purpose of this study.
Cycloplegic retinoscopy was measured using a streak retinoscope (Welch Allyn, Skaneateles Falls, NY) during a preoperative office visit, during which a complete ophthalmic examination was also performed. Cycloplegia was induced with cyclopentolate hydrochloride 1% (Alcon Laboratories, Fort Worth, TX) and tropicamide hydrochloride 1% (Falcon Pharmaceuticals, Fort Worth, TX) eyedrops, instilled 2 to 3 minutes apart. After 30 to 40 minutes, the senior author (DJS) performed cycloplegic retinoscopy with loose trial lenses held approximately 13 mm in front of the eye (vertex distance); the working distance was measured to be 65 cm.
A standard protocol for general anesthesia in children was used throughout this study. Patients were premedicated with oral midazolam if deemed appropriate by the anesthesiologist. Sevoflurane in an oxygen/nitrous oxide mixture was given via facemask for inhalation induction of anesthesia. Intravenous access was obtained, and propofol was given to facilitate intubation. Succinylcholine was administered only when it was necessary to treat laryngospasm during intubation. Anesthesia was maintained throughout the procedure with sevoflurane or isoflurane, with or without nitrous oxide. The depth of anesthesia was titrated based on heart rate, blood pressure and lack of patient movement so as to maintain surgical anesthesia (stage 3). Retinoscopy measurements were performed on subjects under surgical anesthesia. Intraoperative analgesia was maintained with intravenous fentanyl or morphine. Nausea was controlled by intravenous ondansetron alone or in combination with dexamethasone. Intravenous fluids were administered as necessary. Anesthetics used were recorded in the study’s data collection form (data not shown).
Retinoscopy Under General Anesthesia
One eye of each subject was used as the study eye. The study eye received phenylephrine 2.5% eyedrops twice, 2 to 3 minutes apart to dilate the pupil in order to facilitate retinoscopy without inducing cycloplegia. If a unilateral procedure was performed, the fellow eye was used as the study eye. In bilateral procedures, the second eye (most commonly, the left eye) was used as the study eye. The senior author (DJS) performed the measurements using the same technique for streak retinoscopy and the same retinoscope model that was used for cycloplegic retinoscopy. A speculum was not used; the cornea was moistened with balanced salt solution as necessary. An effort was made to use the same vertex distance and working distance as for cycloplegic retinoscopy and to perform the measurement on the visual axis. To minimize examiner bias, the results of cycloplegic retinoscopy were not reviewed before retinoscopy under anesthesia was measured.
The primary outcome of this study was the difference between retinoscopy under anesthesia and cycloplegic retinoscopy for three components of the refractive error: (1) sphere, (2) cylinder and (3) cylinder axis. Bland-Altman analysis was used to assess the level of agreement between the two measurements in order to determine whether retinoscopy under anesthesia and cycloplegic retinoscopy were equivalent. Vector analysis for the cylinder was not performed because the cylinder axis of retinoscopy under anesthesia was within 10 degrees of cycloplegic retinoscopy in all but one subject (see Results). A paired t test was used to compare the difference between the two measurements. Using analysis of covariance (ANCOVA), the effects of age, sex and laterality on the difference between the two measurements was examined. Statistical analysis was performed using SAS software, v 9.2 (Statistical Analysis Systems, Cary, NC).
We enrolled 45 subjects. One subject was excluded from analysis because he had received atropine systemically during anesthesia; one was excluded because she did not match study inclusion criteria (age 18 years); and two others were excluded because a measurement under general anesthesia could not be obtained. Of the remaining 41 subjects, the average age was 3.7 years (SD 2.7; range, 0.8 to 11 years), 27 (65.8%) were female; 11 right eyes (26.8%) and 30 left eyes (73.2%) were included in the study. The diagnoses for which general anesthesia was performed included strabismus (19 subjects), nasolacrimal duct obstruction (15 subjects) and unilateral cataract (2 subjects); limbal dermoid, glaucoma suspect, unilateral glaucoma associated with nevus flammeus, and chalazion were recorded in one subject each, it was not recorded in one. The time between cycloplegic retinoscopy and retinoscopy under anesthesia was less than 6 months for all subjects.
Comparison of Measurements
For the sphere, retinoscopy under anesthesia significantly underestimated hyperopia by 0.98 D compared with cycloplegic retinoscopy (P < 0.0001), but it did provide comparable measurements for the cylinder power and axis ( Table 1 ). For the sphere, retinoscopy under anesthesia was within 0.50 D of cycloplegic retinoscopy in 19 (46.3%) subjects; within 1 D in 25 (61%) subjects; and within 2 D in 36 subjects (87.8%). For the cylinder, it was within 0.50 D in 37 subjects (90.2%) and within 1 D in all subjects. For cylinder axis, retinoscopy under anesthesia was within 5 degrees of cycloplegic retinoscopy in 38 subjects (92.7%) and within 10 degrees in 40 subjects (97.6%); in one subject, the cylinder axis differed by 90 degrees (cylinder power 0.50 D). For the spherical equivalent, retinoscopy under anesthesia was within 0.50 D of cycloplegic retinoscopy in 19 (46.3%) subjects; within 1 D in 28 subjects (68.3%); and within 2 D in 36 subjects (87.8%). Bland-Altman plots show the agreement between retinoscopy under anesthesia and cycloplegic retinoscopy for the sphere ( Figure 1 ); the cylinder ( Figure 2 ); and the spherical equivalent ( Figure 3 ). Bland-Altman analysis for the cylinder axis was not performed because this parameter was not normally distributed.
|Variable||RNS/CY (SD)||RNS/GA (SD)||Mean Difference of RNS/GA, RNS/CY||95% LOA||P value|
|Sphere||1.52 (2.01)||0.53 (2.48)||−0.98 (1.03)||(−3.08, 1.10)||<0.0001|
|Cylinder||−0.44 (0.60)||−0.36 (0.51)||0.08 (0.37)||(−0.67, 0.82)||0.18|
|Spherical Equivalent||1.30 (2.03)||0.37 (2.46)||−0.94 (1.04)||(−3.01, 1.13)||<0.0001|