Analysis of Radiology Report Recommendation Characteristics and Rate of Recommended Action Performance

IMPORTANCE

Following up on recommendations from radiologic findings is important for patient care, but frequently there are failures to carry out these recommendations. The lack of reliable systems to characterize and track completion of actionable radiology report recommendations poses an important patient safety challenge.

OBJECTIVES

To characterize actionable radiology recommendations and, using this taxonomy, track and understand rates of loop closure for radiology recommendations in a primary care setting.

DESIGN, SETTING, AND PARTICIPANTS

Radiology reports in a primary care clinic at a large academic center were redesigned to include actionable recommendations in a separate dedicated field. Manual review of all reports generated from imaging tests ordered between January 1 and December 31, 2018, by primary care physicians that contained actionable recommendations was performed. For this quality improvement study, a taxonomy system that conceptualized recommendations was developed based on 3 domains: (1) what is recommended (eg, repeat a test or perform a different test, specialty referral), (2) specified time frame in which to perform the recommended action, and (3) contingency language qualifying the recommendation. Using this framework, a 2-stage process was used to review patients' records to classify recommendations and determine loop closure rates and factors associated with failure to complete recommended actions. Data analysis was conducted from April to July 2021.

MAIN OUTCOMES AND MEASURES

Radiology recommendations, time frames, and contingencies. Rates of carrying out vs not closing the loop on these recommendations in the recommended time frame were assessed.

RESULTS

A total of 598 radiology reports were identified with structured recommendations: 462 for additional or future radiologic studies and 196 for nonradiologic actions (119 specialty referrals, 47 invasive procedures, and 43 other actions). The overall rate of completed actions (loop closure) within the recommended time frame was 87.4%, with 31 open loop cases rated by quality expert reviewers to pose substantial clinical risks. Factors associated with successful loop closure included (1) absence of accompanying contingency language, (2) shorter recommended time frames, and (3) evidence of direct radiologist communication with the ordering primary care physicians. A clinically significant lack of loop closure was found in approximately 5% of cases.

CONCLUSIONS AND RELEVANCE

The findings of this study suggest that creating structured radiology reports featuring a dedicated recommendations field permits the development of taxonomy to classify such recommendations and determine whether they were carried out. The lack of loop closure suggests the need for more reliable systems.

Artificial intelligence to assess body composition on routine abdominal CT scans and predict mortality in pancreatic cancer- A recipe for your local application

BACKGROUND

Body composition is associated with mortality; however its routine assessment is too time-consuming.

PURPOSE

To demonstrate the value of artificial intelligence (AI) to extract body composition measures from routine studies, we aimed to develop a fully automated AI approach to measure fat and muscles masses, to validate its clinical discriminatory value, and to provide the code, training data and workflow solutions to facilitate its integration into local practice.

METHODS

We developed a neural network that quantified the tissue components at the L3 vertebral body level using data from the Liver Tumor Challenge (LiTS) and a pancreatic cancer cohort. We classified sarcopenia using accepted skeletal muscle index cut-offs and visceral fat based its median value. We used Kaplan Meier curves and Cox regression analysis to assess the association between these measures and mortality.

RESULTS

Applying the algorithm trained on LiTS data to the local cohort yielded good agreement [>0.8 intraclass correlation (ICC)]; when trained on both datasets, it had excellent agreement (>0.9 ICC). The pancreatic cancer cohort had 136 patients (mean age: 67 ± 11 years; 54% women); 15% had sarcopenia; mean visceral fat was 142 cm². Concurrent with prior research, we found a significant association between sarcopenia and mortality [mean survival of 15 ± 12 vs. 22 ± 12 (p < 0.05), adjusted HR of 1.58 (95% CI: 1.03-3.33)] but no association between visceral fat and mortality. The detector analysis took 1 ± 0.5 s.

CONCLUSIONS

AI body composition analysis can provide meaningful imaging biomarkers from routine exams demonstrating AI's ability to further enhance the clinical value of radiology reports.

Multispecialty Enterprise Imaging Workgroup Consensus on Interactive Multimedia Reporting Current State and Road to the Future: HIMSS-SIIM Collaborative White Paper

Diagnostic and evidential static image, video clip, and sound multimedia are captured during routine clinical care in cardiology, dermatology, ophthalmology, pathology, physiatry, radiation oncology, radiology, endoscopic procedural specialties, and other medical disciplines. Providers typically describe the multimedia findings in contemporaneous electronic health record clinical notes or associate a textual interpretative report. Visual communication aids commonly used to connect, synthesize, and supplement multimedia and descriptive text outside medicine remain technically challenging to integrate into patient care. Such beneficial interactive elements may include hyperlinks between text, multimedia elements, alphanumeric and geometric annotations, tables, graphs, timelines, diagrams, anatomic maps, and hyperlinks to external educational references that patients or provider consumers may find valuable. This HIMSS-SIIM Enterprise Imaging Community workgroup white paper outlines the current and desired clinical future state of interactive multimedia reporting (IMR). The workgroup adopted a consensus definition of IMR as “interactive medical documentation that combines clinical images, videos, sound, imaging metadata, and/or image annotations with text, typographic emphases, tables, graphs, event timelines, anatomic maps, hyperlinks, and/or educational resources to optimize communication between medical professionals, and between medical professionals and their patients.” This white paper also serves as a precursor for future efforts toward solving technical issues impeding routine interactive multimedia report creation and ingestion into electronic health records.

Do as AI say: susceptibility in deployment of clinical decision-aids

Artificial intelligence (AI) models for decision support have been developed for clinical settings such as radiology, but little work evaluates the potential impact of such systems. In this study, physicians received chest X-rays and diagnostic advice, some of which was inaccurate, and were asked to evaluate advice quality and make diagnoses. All advice was generated by human experts, but some was labeled as coming from an AI system. As a group, radiologists rated advice as lower quality when it appeared to come from an AI system; physicians with less task-expertise did not. Diagnostic accuracy was significantly worse when participants received inaccurate advice, regardless of the purported source. This work raises important considerations for how advice, AI and non-AI, should be deployed in clinical environments.

Deep Learning to Quantify Pulmonary Edema in Chest Radiographs

Purpose

To develop a machine learning model to classify the severity grades of pulmonary edema on chest radiographs.

Materials and Methods

In this retrospective study, 369 071 chest radiographs and associated radiology reports from 64 581 patients (mean age, 51.71 years; 54.51% women) from the MIMIC-CXR chest radiograph dataset were included. This dataset was split into patients with and without congestive heart failure (CHF). Pulmonary edema severity labels from the associated radiology reports were extracted from patients with CHF as four different ordinal levels: 0, no edema; 1, vascular congestion; 2, interstitial edema; and 3, alveolar edema. Deep learning models were developed using two approaches: a semisupervised model using a variational autoencoder and a pretrained supervised learning model using a dense neural network. Receiver operating characteristic curve analysis was performed on both models.

Results

The area under the receiver operating characteristic curve (AUC) for differentiating alveolar edema from no edema was 0.99 for the semisupervised model and 0.87 for the pretrained models. Performance of the algorithm was inversely related to the difficulty in categorizing milder states of pulmonary edema (shown as AUCs for semisupervised model and pretrained model, respectively): 2 versus 0, 0.88 and 0.81; 1 versus 0, 0.79 and 0.66; 3 versus 1, 0.93 and 0.82; 2 versus 1, 0.69 and 0.73; and 3 versus 2, 0.88 and 0.63.

Conclusion

Deep learning models were trained on a large chest radiograph dataset and could grade the severity of pulmonary edema on chest radiographs with high performance.Supplemental material is available for this article.See also the commentary by Auffermann in this issue.© RSNA, 2021.

Analysis of Kinematic Differences in Hand Motion between Novice and Experienced Operators in IR: A Pilot Study

PURPOSE

To prospectively validate electromagnetic hand motion tracking in interventional radiology to detect differences in operator experience using simulation.

METHODS

Sheath task: Six attending interventional radiologists (experts) and 6 radiology trainees (trainees) placed a wire through a sheath and performed a "pin-pull" maneuver, while an electromagnetic motion detection system recorded the hand motion. Radial task: Eight experts and 12 trainees performed palpatory radial artery access task on a radial access simulator. The trainees repeated the task with the nondominant hand. The experts were classified by their most frequent radial artery access technique as having either palpatory, ultrasound, or overall limited experience. The time, path length, and number of movements were calculated. Mann-Whitney U tests were used to compare the groups, and P < .05 was considered significant.

RESULTS

Sheath task: The experts took less time, had shorter path lengths, and used fewer movements than the trainees (11.7 seconds ± 3.3 vs 19.7 seconds ± 6.5, P < .01; 1.1 m ± 0.3 vs 1.4 m ± 0.4, P < .01; and 19.5 movements ± 8.5 vs 31.0 movements ± 8.0, P < .01, respectively). Radial task: The experts took less time, had shorter path lengths, and used fewer movements than the trainees (24.2 seconds ± 10.6 vs 33.1 seconds ± 16.9, P < .01; 2.0 m ± 0.5 vs 3.0 m ± 1.9, P < .001; and 36.5 movements ± 15.0 vs 54.5 movements ± 28.0, P < .001, respectively). The trainees had a shorter path length for their dominant hand than their nondominant hand (3.0 m ± 1.9 vs 3.5 m ± 1.9, P < .05). The expert palpatory group had a shorter path length than the ultrasound and limited experience groups (1.8 m ± 0.4 vs 2.0 m ± 0.4 and 2.3 m ± 1.2, respectively, P < .05).

CONCLUSIONS

Electromagnetic hand motion tracking can differentiate between the expert and trainee operators for simulated interventional tasks.

Point of Care Image Capture with a Custom Smartphone Application: Experience with an Encounter-Based Workflow

Medical documentation is one of the primary methods by which physicians share clinical information and impressions over time with one another. As the adage says, "a picture is worth a thousand words," and physicians have started leveraging consumer mobile technology to share images with one another. However, image sharing often uses short message service texting and similar methods, which can be noncompliant with privacy regulations and can also limit the ability to communicate information longitudinally and across specialties. Sharing of such information is increasingly important, however, as smaller practices are joining to create large geographically spread out health care networks. To this end, we developed an application to acquire and store images via smartphone and seamlessly transfer into the patient's electronic medical record (EMR) to enable digital consults and longitudinal evaluation in a private and compliant method.

MIMIC-CXR, a de-identified publicly available database of chest radiographs with free-text reports

Chest radiography is an extremely powerful imaging modality, allowing for a detailed inspection of a patient's chest, but requires specialized training for proper interpretation. With the advent of high performance general purpose computer vision algorithms, the accurate automated analysis of chest radiographs is becoming increasingly of interest to researchers. Here we describe MIMIC-CXR, a large dataset of 227,835 imaging studies for 65,379 patients presenting to the Beth Israel Deaconess Medical Center Emergency Department between 2011-2016. Each imaging study can contain one or more images, usually a frontal view and a lateral view. A total of 377,110 images are available in the dataset. Studies are made available with a semi-structured free-text radiology report that describes the radiological findings of the images, written by a practicing radiologist contemporaneously during routine clinical care. All images and reports have been de-identified to protect patient privacy. The dataset is made freely available to facilitate and encourage a wide range of research in computer vision, natural language processing, and clinical data mining.

Migrating to the Modern PACS: Challenges and Opportunities

With progressive advancements in picture archiving and communication system (PACS) technology, radiology practices frequently look toward system upgrades and replacements to further improve efficiency and capabilities. The transition between PACS has the potential to derail the operations of a radiology department. Careful planning and attention to detail from radiology informatics leaders are imperative to ensure a smooth transition. This article is a review of the architecture of a modern PACS, highlighting areas of recent innovation. Key considerations for planning a PACS migration and important issues to consider in data migration, change management, and business continuity are discussed. Beyond the technical aspects of a PACS migration, the human factors to consider when managing the cultural change that accompanies a new informatics tool and the keys to success when managing technical failures are explored. Online supplemental material is available for this article. ^©RSNA, 2018.

Resident iPad use: has it really changed the game?

PURPOSE

The purpose of this study was to assess residents' usage patterns and opinions of the iPad as a tool for radiology education and clinical practice at an academic medical center.

METHODS

All 38 radiology residents in our radiology program (postgraduate years 2 to 5) were provided with iPad 2 tablets and subscriptions to e-Anatomy and STATdx. After 6 months of device use, residents were surveyed to assess their opinions regarding the technology as a tool for education and clinical practice.

RESULTS

A total of 36 residents (95%) completed the survey. Eighty-six percent reported daily iPad use. Radiology-specific applications, particularly e-Anatomy, were used weekly or daily by 88% of respondents. Most preferred to read journal articles on the iPad (70%), but the number of respondents preferring to read textbooks on the iPad (48.5%) compared with the traditional bound form (48.5%) was evenly divided. Residents were also divided on the clinical utility of the iPad. Most had not used the iPad to view radiologic examinations (75%). Fewer than half (47%) used their iPads during readout. Finally, only 12% had used the iPad to edit dictated reports.

CONCLUSIONS

The iPad has generated excitement within the radiology community, particularly among resident educators, who are increasingly recognizing the unique needs of "millennial learners." This study showed that the majority of residents at the authors' institution have incorporated the iPad as an educational tool and use it as a learning aid. Incorporation of the iPad into clinical workflow has been less pronounced.

Early preventive interventions for adolescents exposed to a potentially traumatic event

The development and evaluation of early preventative interventions for individuals exposed to a potentially traumatic event (PTE) have been increasing in number and sophistication over the last decade. Evaluation of early interventions for individuals who are symptomatic several weeks after exposure to a PTE have been effective in preventing the development of chronic post-traumatic stress disorder (PTSD). However, there has been little research demonstrating effectiveness of immediate interventions. In addition, there are not specific early interventions targeted toward the adolescent age group. While providers of early interventions should initiate contact with the family as whole, they may need to decide whether to employ a model that was developed for both children and adolescents or for adults depending on the particular needs of the adolescent. This review will examine a range of issues regarding the meaning, timing and effectiveness of early interventions after a PTE and will cover models for all age groups as well as models designed for immediate crisis intervention as well as for those individuals who are symptomatic and distressed in the early post-traumatic period. The rapid growth in the formulation and research of early intervention models and the success of certain models in the prevention of chronic PTSD are promising developments for establishing an evidence base of cost effective preventative early interventions.

Prevalence of left ventricular regional dysfunction in arrhythmogenic right ventricular dysplasia: a tagged MRI study

BACKGROUND

Although arrhythmogenic right ventricular dysplasia (ARVD) predominantly affects the right ventricle (RV), genetic/molecular and histological changes are biventricular. Regional left ventricular (LV) function has not been systematically studied in ARVD.

METHODS AND RESULTS

The study population included 21 patients with suspected ARVD who underwent evaluation with MRI including tagging. Eleven healthy volunteers served as control subjects. Peak systolic regional circumferential strain (Ecc, %) was calculated by harmonic phase from tagged MRI based on the 16-segment model. Patients who met ARVD Task Force criteria were classified as definite ARVD, whereas patients with a positive family history who had 1 additional minor criterion and patients without a family history with 1 major or 2 minor criteria were classified as probable ARVD. Of the 21 ARVD subjects, 11 had definite ARVD and 10 had probable ARVD. Compared with control subjects, probable ARVD patients had similar RV ejection fraction (58.9+/-6.2% versus 53.5+/-7.6%, P=0.20), but definite ARVD patients had significantly reduced RV ejection fraction (58.9+/-6.2% versus 45.2+/-6.0%, P=0.001). LV ejection fraction was similar in all 3 groups. Compared with control subjects, peak systolic Ecc was significantly less negative in 6 of 16 (37.5%) segments in definite ARVD and 3 of 16 segments (18.7%) in probable ARVD (all P<0.05).

CONCLUSIONS

ARVD is associated with regional LV dysfunction, which appears to parallel degree of RV dysfunction. Further large studies are needed to validate this finding and to better define implications of subclinical segmental LV dysfunction.

Axial black blood turbo spin echo imaging of the right ventricle

Black blood turbo spin echo (TSE) imaging of the right ventricle (RV) free wall is highly sensitive to cardiac motion, frequently resulting in nondiagnostic images. Temporal and spatial parameters of a black blood TSE pulse sequence were evaluated for visualization of the RV free wall. Seventy-four patient studies were retrospectively evaluated for the effects of acquisition timing on image quality. Axial black blood TSE images were acquired on 10 healthy volunteers to assess the role of spatial misregistration on right ventricle visualization; increasing the double inversion recovery (DIR) slice thickness beyond 300% had no effect on image quality (P = 0.2). Thirty-five patient studies were prospectively evaluated with inversion times (TIs) corresponding to the mid-diastolic rest period and end-systole based on visual analysis of a four chamber cine. When TIs were chosen to be within the patients' RV rest period, mean image quality score was significantly improved (2.3 vs 1.86; P < 0.001) and the number of clinically diagnostic images increased from 32% to 46%. Black blood TSE imaging of the RV free wall is highly sensitive to cardiac motion. Image quality can be improved by choosing TIs concordant with the rest period of the patient's RV that may occur at mid-diastole or end-systole.

Fatal nonneurological EHV-1 infection in a yearling filly

A case of fatal nonneurological equine herpesvirus 1 (EHV-1) infection in a yearling filly is described. Gross lesions included extensive pulmonary edema, prominent laryngeal lymphoid follicles, and congestion and edema of the dorsal third ventricle choroid plexus. Histologically, there was vasculitis, hemorrhage, and edema in the lungs and dorsal third ventricle choroid plexus as well as mild intestinal crypt necrosis with occasional intranuclear inclusion bodies. The perivascular and vascular inflammatory infiltrates were comprised mainly of T lymphocytes and macrophages. EHV-1 antigen was identified within the nucleus and cytoplasm of endothelial cells, dendritic-like cells of the pharyngeal lymphoid follicles, pharyngeal glandular epithelium, crypt enterocytes, and monocytes. Attempted virus isolation was negative. Weak seroconversion for EHV-1 was observed. Herpesvirus-like particles were identified within pharyngeal endothelial cells by transmission electron microscopy. Polymerase chain reaction amplified 369 and 188 base-pair fragments specific for EHV-1. The scarcity of pathognomonic viral inclusions and lesions in this case suggests that this disease may not be recognized, particularly in situations when ancillary laboratory procedures are limited.

The Child Development-Community Policing Program: a partnership to address the impact of violence

This article describes the development, evolution and current status of the New Haven Child Development-Community Policing (CDCP) Program, a partnership between child mental health professionals at the Yale Child Study Center and the New Haven Department of Police Service. The central mission of the program is to intervene early in an attempt to ameliorate the effects of children's exposure to violence. Recent programmatic expansion of the CDCP program into areas of juvenile offenders, specialized protocols for children affected by domestic violence, and school based groups are also explained. There are five core elements of the CDCP program: (1) Seminars in applied child development for police officers; (2) Fellowships for advanced police officers; (3) Fellowships for clinicians who are trained both in seminars and experientially in police practice and procedure; (4) A weekly program conference; and (5) A clinical consultation service in which clinicians are available 24 hours per day, 7 days a week are explicated.

Police and mental health professionals. Collaborative responses to the impact of violence on children and families

Coordinating responses through the Child Development-Community Policing Program has led to multiple changes in the delivery of clinical and police services. Mental health clinicians and police officers have developed a common language for assessing and responding to the needs of children and families who have been exposed to or involved in violence. Learning from each other, these unlikely partners have established close working relationships that improve and expand the range of interventions they are able to provide while preserving the areas of expertise and responsibilities of each professional group. The immediate access to witnesses, victims, and perpetrators of violent crimes through the consultation service provides a unique opportunity to expand the understanding of clinical phenomena from the acute traumatic moment to longer-term adaptation, symptom formation, and recovery. In turn, the initiative introduces the systematic study of basic psychological and neurobiologic functions involved in traumatization as well as the investigation of psychotherapeutic and pharmacologic therapies. Similarly, program involvement with juvenile offenders has led to a coordinated response from the police, mental health, and juvenile justice systems. This project provides an opportunity to develop detailed psychological profiles and typologies of children engaged in different levels of antisocial behavior as well as to determine the characteristics that might predict with whom community-based interventions might be most successful. A recent survey of New Haven public school students has yielded promising evidence that community policing and the program are having a positive impact on the quality of life. In a survey of sixth-, eighth-, and tenth-grade students there were substantial improvements in students' sense of safety and experience of violence between 1992 and 1996. When asked if they felt safe in their neighborhood, there was an increase in the percentage of positive responses from 57% to 62% for sixth-grade students, 48% to 66% for eighth-grade students, and 53% to 73% for tenth-grade students, and when asked if they had seen someone shot or stabbed there was a decrease in positive responses from 43% to 28% for sixth-grade students, 46% to 31% for eighth-grade students, and 34% to 28% for tenth-grade students. Today, we are all too familiar with the developmental trajectory that leads children into violent crime. Newspaper articles and clinical case reports have taken on a dreary repetitiveness. These young criminals are often poor, minority, inner-city children who are known to many agencies to be at risk because of family disorganization, neglect, and abuse. They are failing in school or are already on the streets. One day they are victims and the very next they are assailants. We are all familiar with the inadequacies in the social response to these children, from their preschool years through the point at which they become assailants themselves. What is shocking is that the age at which children make the transition from being abused to being abusive seems to be getting earlier, and the number appears to be increasing. On the positive side, there is an increased awareness of the need and the ability of the various sectors of society to respond in concert. The institutions that function in the inner city--schools, police, mental health and child welfare agencies, churches--are all concerned about the same children and families. By working together, with a shared orientation to the best interests of the children, they can intervene earlier and more effectively: first, to disrupt the trajectory leading to violence; and, second, to help those children who are already caught in the web of exposure to violent crime and inner-city trauma. The experience with community-based policing and mental health in New Haven, now being replicated throughout the United States, can thus stand as a model of an active social response to an overwhel

Intensive, integrated, in-home psychiatric services. The catalyst to enhancing outpatient intervention

The authors introduce the Yale Intensive In-Home Child and Adolescent Psychiatric Service, a model of home-based care for children with severe psychiatric disturbances. This model synthesizes the principles and method of the wrap-around paradigm and in-patient child psychiatric practice within the reality of the managed care system. A clinical team, under the direct supervision of a child psychiatrist, works directly within the family to understand and address the multilevel transactions that have affected the child's ability to function in various domains and resulted in recommendations for intensive intervention, including psychiatric hospitalization. This article suggests that if the psychiatrist is to provide the highest level of care, cognizance of and involvement in the child's ecology are as essential for the child and adolescent psychiatrist as other aspects of the child's world and life. In the days of ever shortening patient lengths of stay, this model of care offers promise for both clinical and fiscal effectiveness.

The Computerized Laboratory Notebook concept for genetic toxicology experimentation and testing

We describe a microcomputer system utilizing the Computerized Laboratory Notebook (CLN) concept developed in our laboratory for the purpose of automating the Battery of Leukocyte Tests (BLT). The BLT was designed to evaluate blood specimens for toxic, immunotoxic, and genotoxic effects after in vivo exposure to putative mutagens. A system was developed with the advantages of low cost, limited spatial requirements, ease of use for personnel inexperienced with computers, and applicability to specific testing yet flexibility for experimentation. This system eliminates cumbersome record keeping and repetitive analysis inherent in genetic toxicology bioassays. Statistical analysis of the vast quantity of data produced by the BLT would not be feasible without a central database. Our central database is maintained by an integrated package which we have adapted to develop the CLN. The clonal assay of lymphocyte mutagenesis (CALM) section of the CLN is demonstrated. PC-Slaves expand the microcomputer to multiple workstations so that our computerized notebook can be used next to a hood while other work is done in an office and instrument room simultaneously. Communication with peripheral instruments is an indispensable part of many laboratory operations, and we present a representative program, written to acquire and analyze CALM data, for communicating with both a liquid scintillation counter and an ELISA plate reader. In conclusion we discuss how our computer system could easily be adapted to the needs of other laboratories.

Vitamin A nutrition of the equine: growth, serum biochemistry and hematology

Four diets were fed to pony fillies for 40 weeks. One group received a basal diet low in carotene, designated mildly deficient. Other groups were fed basal diet plus vitamin A propionate equivalent to 12 (control), 1,200 (mildly intoxicated) or 12,000 (severely intoxicated) microgram retinol/kg body weight/day. The mildly deficient group exhibited impairment of growth and hematopoiesis with decreased serum concentrations of iron, albumin and cholesterol prior to depletion of liver vitamin A. Growth was depressed in mildly intoxicated and severely intoxicated fillies. The latter became debilitated and died in less than 40 weeks with reduced serum concentrations of iron, albumin and cholesterol. Significant associations were calculated between dietary vitamin A and vitamin A concentrations in plasma, liver and kidney, and between plasma vitamin A and liver and kidney vitamin A. Plasma retinyl ester increased logarithmically with plasma total vitamin A and linearly with kidney vitamin A. The data suggest that increases in retinyl ester due to excessive dietary vitamin A occur prior to appearance of clinical signs and perhaps reflect kidney vitamin A levels. Calculated estimates of optimal vitamin A intake fo maximal growth averaged 1.4 times the current National Research Council recommendation, for liver secreted serum constituents 5.4 times and for red blood cell criteria 10 times. Thus the recommended intake of 12 microgram/kg for weanling horses appears less than optimal. The data suggest hematologic criteria and serum biochemistries may be more sensitive indicators of vitamin A nutriture than growth.