Click for next page ( 2


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 1
Chapter I ORGANIZATION AND CONDUCT OF STUDY Gilbert W. Beebe and Barnes Woodhall A. INTRODUCTION At the time of the Conference on Postwar Research called by the Na- tional Research Council (NRC) (56) in 1946, prominent reference was made to the desirability of a study of peripheral nerve injuries based on the Army Registry, and, when the NRC accepted from the Armed Forces and the Veterans Administration (VA) a mandate to develop the means for exploiting the medical experience of World War II, it was inevitable that the present study be among the first proposed. A preliminary plan was circulated by Dr. Woodhall in the fall of 1946 and at the request of the VA the NRC called, in January 1947, a Peripheral Nerve Conference 8 to review objectives, prepare specific plans, and settle upon the centers where necessary work might be performed. The resulting research protocol, further refined by those who were to take the lead in the conduct of the subsequent work, was submitted to the VA through the Committee on Veterans Medical Problems of the NRC and recommended for financing in March. 1947 in the first group of projects under the new program. As one of the first studies sponsored by the Committee, the peripheral nerve study was begun without having the benefit of any prior experience in medical follow-up studies on veterans, or of existing procedures for insur- ing access to veterans and their records. It was, in fact, in large part through the peripheral nerve project that the procedures necessary to the success of the entire VA-NRC follow-up program were evolved. The research plan was a very general one, concerned chiefly with the specification of the follow-up observations to be made and of the university * Those attending were: Conference Members—Drs. Loyal Davis (Chairman in Dr. Woodhall's unavoidable absence), Harry Grundfcst, Thomas Hoen, Frederic H. Lcwey, W. K. Livingston, Lewis J. Pollock, Curt P. Richter, R. G. Spurting, W. P. Van Wagenen, Arthur J. Watkins, Paul W. Weiss, and James C. White. V. S. Army—in. Col. R. P. Mason, M. C., War Department General Staff. Major S. J. Vogel, Jr., M. C., Office of The Surgeon General. Mr. J. J. Ozog, Medical Statistics Division, Office of The Surgeon General. U. S. Ncay—Captain George B. Dowling (MC), Bureau of Medicine and Surgery. Commander R. A. Phillips (MC), Bureau of Medicine and Surgery. Lt. (j. g.) E. N. Weaver (MC), Naval Hospital, Bethesda, Md. Veterans Administration—Drs. E. H. Cushing and C. Harrison. Motional Research Coutuil—Vrs. Lewis H. Weed, S. D. Aberle, Gilbert W. Bcebe, and John C. Ransmeier.

OCR for page 1
centers where the patients would be seen Before actual work could begin it was necessary for the investigators to engage in more definitive planning both with respect to the observations to be made and the selection of cases for study, and for the Committee and the VA to work out procedures for facilitating access to both records and patients. It was also necessary to develop an administrative pattern for the organization of the work as a cooperative endeavor of the investigators on the one hand and the Com- mittee, through its Follow-up Agency,7 on the other. Specific planning along these lines proceeded only slowly and it was a year before any real beginning could be made on the work. Beginning in the January 1947 Peripheral Nerve Conference and con- tinuing thereafter, great emphasis was placed upon the prospective value of detailed, largely quantitative studies of specific modalities in contrast to broad, functional assessments, and the observations eventually decided upon at a final planning conference at Hot Springs, Va., in November 1947 reflect the former point of view. Also, since greatest emphasis was placed upon complete lesions, the Army Registry was visualized as the chief source of case material, but sufficient interest was expressed in neu- rolysed lesions and certain other groups of special interest to require the acquisition of other rosters as well. From the outset, and continuing throughout the life of the project, finan- cial support has come entirely from the VA and has been generous in its extent. The Armed Forces have furnished the rosters and basic clinical data upon which the entire study rests, as well as strategic aid of other kinds. Support by the VA has not been confined to the financing of the project, but has extended on the provision of information essential to the location and follow-up of the individuals to be studied, and to many other services. B. ORGANIZATION OF STUDY There are three elements in the organization of the actual work of the project: (1) The chairman, Dr. Barnes Wood hall; (2) the 5 follow-up centers headed by Dr. James C. White of Boston, Dr. Loyal Davis of Chicago Dr. Harry Grundfest of New York, Dr. Frederic H. Lewey (deceased) of Philadelphia, and Dr. Howard C. Naffziger of San Francisco; and (3) the NRC Follow-up Agency. The association was a loose, voluntary one, each center and the NRC Follow-up Agency being financed by means of an independent research contract with the VA. The chairman, at the same time Peripheral Nerve Consultant to the VA, assumed responsibility for the sampling plan, the allocation of patients for study, the development of procedures for facilitating access to patients, the provision of medical records covering the original injury and its treatment, assisting the centers in establishing a standard protocol for the follow-up examination, and for the final collection, analysis, and publication of the basic data obtained. 7 Under the Committee there was established the Follow-up Agency as a records and statistical organization to provide operating assistance to investigators participating in the entire VA-NRC follow-up program.

OCR for page 1
In most of these functions the chairman had the assistance of the NRC Follow-up Agency established under the Committee on Veterans Medical Problems to assist investigators in obtaining necessary access to patients and their records, and in processing their observations, not merely for the peripheral nerve project but for any which might be sponsored by the Committee. The senior author of the present chapter, a statistician, was assigned to the project to assist the chairman and directors of follow-up centers in these aspects of the study. Selection of the follow-up centers was the major step taken in organizing the study. It reflects in part the organization of clinical work in the Army during World War II with its emphasis upon centers for specialized care, and wartime support of research directed toward the better diagnosis of peripheral nerve lesions. Even prior to the Peripheral Nerve Conference in January 1947, four investigators in appropriately large urban centers (Boston, Chicago, New York, and Philadelphia) had indicated to Dr. Woodhall their desire to participate in a precise and detailed assessment of the final level of regeneration and functional return following injury of known extent. The desirability of a West Coast center was stressed at the January meeting, and a fifth investigator, Dr. Naffziger, was invited to participate in the planning of the project with a view to establishing a center in San Francisco. Dr. Naffziger accepted the invitation and the final recommendation of the Conference on Peripheral Nerve Injuries was that there be five centers located as mentioned. This recommendation was based on no specific consideration of sample size required by the objectives of the investigation, but the present investigation was entirely too complex for an optimum sample size to be calculated on the basis of purely statistical considerations. The consensus of the conference was that an adequate study would require perhaps several thousand cases. Each follow-up center was organized somewhat differently in terms of the number and variety of professional personnel assisting the responsible in- vestigator in the work. At the January 1947 conference a beginning was made on the drafting of a protocol governing the follow-up examinations, and in their formal applications to the VA for research contracts, the investigators bound themselves to develop and observe such a protocol. In addition, it was expected that each investigator would supplement the standard observations with others, relevant to the project, in which his group had special interest and competence. Also, the great emphasis upon detailed and objective observations, as distinguished from summary ratings of functional return, tended to shape the organization in terms of specialists in several of the centers. At the Chicago center, for instance, where most patients were hospitalized for 4 or 5 days while their studies were being made, primary responsibility for different elements of the examination was assigned in terms of sensory examination, motor exam- ination, electrical studies, sympathetic function studies, photography, and a final surgical evaluation and interpretation of regeneration by the re- sponsible investigator. In other centers patients were examined on an

OCR for page 1
ambulatory basis, usually over a 2-day period, but the same concept of specialized examinations prevailed, although specific interests might vary from center to center. Neuropathological studies were done under a separate contract with Duke University, where Dr. William R. Lyons worked with the chairman on the classification of nerve ends for eventual correlation with the results of the clinical follow-up studies. Allocations of patients for study were made by the chairman through the NRC Follow-up Agency, and at the same time photostatic copies of all relevant medical records were provided to all but the Chicago center, where necessary data were abstracted from original records in VA claims folders. Allocations were made on standard NRC forms bearing the latest address obtainable by the Follow-up Agency, usually through the local VA office. It then became the responsibility of the center to arrange for the examination and to record the results on special forms worked out in conference by representatives of all centers. About midway through the study a statistical code was developed to cover all the data in the stand- ard protocol. The code was divided into two parts, corresponding to the military and civilian periods. Coding based on the first part, dealing with the injury and its management, was performed centrally by personnel of the Follow-up Agency trained and supervised by the chairman and the project statistician. The second part of the coding, covering the interim history and the follow-up examination, was done in each center under the direction of the responsible investigator. All eligible cases were coded, whether examined or not, and the coded information deposited with the NRC Follow-up Agency for editing, tabulation, and statistical analysis in accordance with plans developed by the investigator responsible for the particular subject, assisted by the project statistician. The latter functions, and that of preparing manuscript, were divided among the investigators in November 1950 in accordance with a working outline, as follows: Responsibility __ Chapter and subject Primary Secondary I. Introduction Woodhall. II. Organization and conduct of Beebe Woodhall. study. III. Characteristics of the sample.... Woodhall Beebe. IV. Motor recovery Yahr Nulsen, Herz. V. Electrodiagnosis Grundfest Davis, Moldaver, Webb. VI. Sensory recovery DavU White (pain), Nulsen* Herz. VII. Autonomic regeneration White Herz. VIII. Pathology Lyons Woodhall. IX. Summary A. Correlation of modalities. .. Yahr. B. Anatomic regeneration Davis, Grundfest, Yahr, White. C. Brachial plexus cases Nulsen Davis. D. Functional return Nulsen.

OCR for page 1
Responsibility Chapter and subject Primary Secondary IX. Summary—Continue! E. Symptoms, treatment, and Nulsen. adjustment. X. Neurosurgical implications Woodhall Davis, White. XI. Program for management of per- Woodhall Nulsen, Davis, White. ipheral nerve injuries. The final table of contents bears only an approximate relationship to this original working outline. C. SAMPLING PLAN AND ALLOCATION OF CASES FOR STUDY The research protocol leading to acceptance of the study by the Commit- tee and the VA contains only a general statement of objectives: "The primary purpose of this study is that of evaluating the results of peripheral nerve injuries sustained in World War II* with the hope of standardizing such treat- ment for future wars and* where possible, for similar injuries of civilian life. The secondary purpose of this study, and one of considerable immediate value, is that of discovering nerve injuries among veterans of all services that still require remedial measures." The specific objectives which dictated the sampling plan, however, may be listed as follows: 1. To describe the final level of regeneration in representative cases of complete suture, neurolysis, and nerve graft. 2. To ascertain the apparent influence of gross characteristics of the lesion, and of associated injuries, upon final result. 3. To ascertain the apparent influence of numerous aspects of management upon final result. 4. To evaluate predictions of final recovery based on gross and histologic study of tissue removed at operation. Many other objectives, of course, moved the investigators to undertake the present study but exert no influence on the sampling plan. The minutes of the various conferences and lesser meetings reveal expressions of interest in the following: case-finding, correlation of careful research examinations with VA disability ratings, improvement in accuracy of VA disability ratings, correlation between different modalities, e. g., motor and sensory, evaluation of techniques of assessing recovery, evaluation of military dis- position policy, evaluation of rehabilitative measures undertaken on World War II cases, and study of psychological factors interfering with return of function commensurate with level of nerve regeneration. No interest was expressed in the population of perhaps 40,000 World War II peripheral nerve injuries as a whole, certainly a most difficult one to specify and, by that token, to sample. The uncertainty extends not merely to lesions with spontaneous recovery but also to those treated by neurolysis because of their nonspecific character. Some care was exercised, therefore, in acquiring rosters of neurolyses. Also, only certain major peripheral nerves and the brachial plexus were of real interest, namely,

OCR for page 1
the median, ulnar, radial, peroneal, tibial, and sciatic. Further, each nerve was viewed as a distinct entity requiring some separate study, with the result that special efforts were required to insure adequate numbers of each. A minor consequence of such sampling, of course, is that the distribution of lesions by nerve provides no real estimate of that in the general population of peripheral nerve injuries. There were also certain factors of special interest because of possible paucity of cases in a representa- tive sample, and it was thought necessary to take special precautions to obtain adequate numbers of cases, e. g., those involving arterial injury, normal bone resection, bulb suture, nerve graft, pathological study of nerve ends by Dr. Lyons and Dr. Woodhall, early suture overseas, and sutures done by the plasma glue technique. In summary, then, the objectives of the study require both adequate, representative samples and groups of cases of special interest for particular comparisons. Rosters of bona fide peripheral nerve injuries were acquired on the basis of the above interests and these merit careful description. Rosters of Complete Lesions Primarily Roster #39, contributed by James C. White, M. D., of Boston, Mass., and consisting of 31 peripheral nerve injuries treated by him at St. Albans Naval Hospital. Roster #48, contributed by Thomas I. Hoen, M. D., of New York City, and representing every nerve repair he did at St. Albans Naval Hospital from July 1943 to June 1946, a series of 143 consecutive cases, all repaired by the plasma glue technique. Roster #68, contributed by The Surgeon General, U. S. Army, and con- sisting of 375 1944 Army wounded who were coded as having wounds or fractures with nerve involvement. This roster was used in estimating the incidence and variety of peripheral nerve injuries and particularly in the tests of the completeness of the Army Peripheral Nerve Registry which are described below. Roster #69, the Army Peripheral Nerve Registry, established by the chairman during World War II and contributed to the study by The Surgeon General, U. S. Army. In November 1944, Z/I neurosurgical centers were directed (82) to register all peripheral nerve injuries requiring suture or graft, or under study following such treatment, and to file reports on the completion of any repair and on reassessment at 3-month intervals and at disposition. As received from The Surgeon General, the roster consists of 7,720 men with one or more nerve injuries, and undoubtedly represents the bulk of the sutured cases in the World War II Army experience. Roster #86, contributed by Donald H. Wrork, M. D., of Rockford, 111., and consisting primarily of peripheral nerve injuries with early suture which he performed at the 15th Evacuation Hospital in the Mediterranean Theater of Operations. The total roster numbers 102 men.

OCR for page 1
Roster #88, contributed by Barnes Woodhall, M. D., of Durham, N. C., and William R. Lyons, M. D., of San Francisco, Calif., and consisting of 555 peripheral nerve injuries, treated at Walter Reed and Halloran General Hospitals, for which pathological material was available at the Armed Forces Institute of Pathology (Accession No. 110822). Most of these cases are duplicated on the Army Peripheral Nerve Registry. Rosters Obtained as Sources of Incomplete Lesions Roster #40, contributed by W. K. Livingston, M. D., of Portland, Oreg., and consisting of 1,278 Navy and Marine Corps wounded men with periph- eral nerve injuries of all kinds managed by Dr. Livingston. Only incom- plete lesions were used in the allocations. Roster #47, contributed by Everett G. Grantham, M. D., of Louisville, Ky., and consisting of 619 men with peripheral nerve injuries treated by himself and Claude Pollard, M. D., at Tilton and England General Hospitals. Only incomplete lesions were used. Roster #90, contributed by the late Frederic H. Lewey, M. D., of Phila- delphia, Pa., and consisting of 1,220 men with peripheral nerve injuries seen by him at Cushing General Hospital. The roster was used primarily as a source of incomplete lesions. Roster #92, contributed by Harry Grundfest, Ph. D., of New York City, and consisting of 305 men with incomplete peripheral nerve injuries treated at Halloran General Hospital. Once acquired, the various rosters were processed to yield essentially two samples, which may be termed "representative" and "extended" in order to distinguish them, each of which was further subdivided into sutures and grafts on the one hand and neurolyses on the other. The distinction between the representative and the extended samples was made on the basis of geographic areas surrounding the follow-up centers and applies only to the Army Registry. Each man appearing on the Registry was located geographically, and for each center appropriate geographic limits were stipulated for each nerve, varying somewhat in- versely in size with the relative frequency of injury to the particular nerves. The specific geographic limits adopted are as follows, by center and by nerve: Boston: Brachial plexus Massachusetts, Rhode Island, Con- necticut, New Hampshire, and Maine. Median 70 mile area around Boston. Peroneal 125 mile area around Boston. Radial 100 mile area around Boston. Sciatic 120 mile area around Boston. Tibial 125 mile area around Boston. Ulnar Massachusetts, 125 miles to the west of Boston, plus Rhode Island.

OCR for page 1
Chicago: Brachial plexus 320 mile area around Chicago. Median Cook County. Peroneal Cook and DuPage counties in Illinois* plus the southern part of Michigan. Radial Cook County. Sciatic Cook County. Tibial Most of Illinois, the northern part of Indiana, and the southern halves of Wisconsin and Michigan. Ulnar Cook County. JV«u Tmk: Brachial plexus All of New York, plus 1 case in New Jersey. Median The 5 boroughs of New York City* plus Essex, Middlesex, and Union counties in New Jersey. Peroneal New York City, plus northern New Jersey and Orange, Putnam, West- chester, and Rockland counties in New York. Radial The 5 boroughs of New York City, plus Hudson County in New Jersey. Sciatic The 5 boroughs of New York City, plus Bergen and Hudson counties in New Jersey. Tibial 60 mile area around New York City. Ulnar The 5 boroughs of New York City. Philadelphia: Brachial plexus Pennsylvania* Maryland, Virginia, and West Virginia. Median The eastern part of Pennsylvania and that part of Maryland and Delaware most accessible to Philadelphia. Peroneal Eastern Pennsylvania and part of Maryland. Radial 140 mile area around Philadelphia. Sciatic 140 mile area around Philadelphia. Tibial 130 mile area around Philadelphia. Ulnar 145 mile area around Philadelphia. San Francisco: Brachial plexus California. Median California. Peroneal California. Radial California. Sciatic California. Tibial California. Ulnar California, plus 1 case in Washoc County, Nev. It will be noted that men back in service are excluded under this plan, and although few they are of some special interest. Accordingly, the chairman made a separate study of 18 men who were in service at the time the allocations were being made; Army surgeons generously undertook to examine these men according to the general specifications sent them under

OCR for page 1
this plan. All the men from the Registry residing in the above areas were allocated as part of the representative sample, and all the sutures (both complete and partial) included in the representative sample were thus obtained. Special efforts were then made to augment the allocation with men living outside these areas but having a particular interest in connection with one or another of the special topics listed above, and such men are not included in the representative sample, either for sutures or for neuroly- ses. Men from other rosters were allocated solely on the basis of propinquity to the nearest center and do not contribute sutures to the representative sample; they do, however, contribute neurolyses to the representative sample. Figure 1 provides a rough diagram of these various interrelations. The shaded areas taken together constitute the representative sample, the total area the extended or total sample. All sutures not from the Registry were acquired with special interests in mind, and, therefore, are possibly atypical, and this is also true of all Registry cases outside the sampling area. Figure 1. Relation Between Representative and Extended Samples Roster and Sampling Area Type of Surgery Suture Neurolysis R ep r e s entati ve Sample Legend Extended Sample Two groups of cases were added locally. Dr. Lewey was especially interested in long-gap cases and those with incomplete lesions, and he conducted some reexaminations at Cushing in connection with these special interests. Any such Cushing cases which appeared on the Army 403930—57-

OCR for page 1
Registry were placed in the basic sample or outside it, depending entirely upon geographical considerations. The other special group consisted of men with plasma glue sutures (roster 48), who were seen only at the New York center. These cases are excluded from the representative sample of sutures but are used at most points in the analysis. Men were allocated on the basis of a single nerve injury, and if a man had more than a single injured nerve the interest in him depended upon the roster to which he belonged. When a man was examined, of course, all his injured nerves were taken into account. Thus, the final allocation of 2,714 men had on the order of 3,800 lesions worthy of study, as such lesions are counted here. By center, the allocation of men was as follows; Boston 378 Chicago 389 New York 849 Philadelphia 746 San Francisco. . . . . 352 Total 2, 714 Table 1 indicates the use made of each roster in preparing the allocations. Table 1.—Distribution of Final Allocations by Roster Number of men allocated Roster number and source 39 (White) 31 40 52 47 (Grantham and Pollard) 74 48 (Hocn) 127 61 12 68 (SCO) 69 (Army Registry) 1,987 86 (Wrork) 33 88 193 90 (Lewey) 183 97 (Grundfest) 22 Total 2,714 1 In addition there were 163 men allocated from roster 69, so that the pathological roster is represented by a total of 356 men in the entire allocation. Two chief assumptions underlie the sampling plan: (1) that the Registry would produce representative cases within any definitive-operation group; and (2) that sampling based on residence is unbiased as to nerve regenera- tion. Only the first of these has seemed worthy of investigation. In view of the care attending the framing of the Army directives establishing the Registry, and the efficient concentration of Army cases in Z/I neuro- 10

OCR for page 1
surgical centers, there could be little doubt about the representativeness of sutures drawn from the Registry, but, nevertheless, a limited test was made on the 1944 material, using roster 68 as the basis of expectation. The procedure consisted of, first, identifying all roster 68 cases having peripheral nerve lesions treated by suture or graft, and second, collating this list against the Registry. There were in all 42 roster 68 men with such sutures or grafts, of which 18 were found on the Registry. The 24 cases excluded from the Registry were then compared with the 18 over- lapping cases, the points of comparison being rank, month of injury, place of injury, place of final treatment, evacuation home, coding of first, second, and third diagnoses, location of first diagnosis, operations performed, final result of injury, disposition from hospital, and total time spent in hospital. In none of these comparisons was a statistically significant (P<.05) difference found. As far as this very limited exploration goes, therefore, it suggests that: (1) about half of the 1944 sutures and grafts appear on the Registry; and (2) there are no very considerable differences between cases included and excluded from the Registry. The bulk of the Registry cases, of course, are of 1945 origin and it was not thought necessary to carry the exploration of representativeness and completeness into the 1945 material. In one minor respect the lysed cases from the Registry are surely atypical: of necessity, each involves a man with at least one other nerve injury, usually on the same limb, treated by suture or graft. Any such bias assumes importance to the extent that the presence of associated nerve lesions might affect recovery, but, fortunately, this is of no real consequence except in assessing practical function, where the influence of associated lesions must be controlled in any event. The lysed cases from the Registry may also be unrepresentative of the generality of neurolyses in the same way in which the rosters of incomplete lesions are probably atypical, i. e., in that the lesions were more carefully evaluated, more skillfully handled, and more surely involved truly denervated muscles. But in this instance one is not interested in the generality of so-called neurolyses but in only the well-studied, well-documented incomplete lesions which these rosters represent. After the reexaminations had been completed, during the coding of the details of injury and management, additional criteria of eligibility were introduced which led to the exclusion of 62 men on the following grounds: Injury occurred before entry into service. Defect not caused by trauma. Injury caused only sensory deficit. Spinal injury. In addition, card-punching of follow-up material coded by the several centers was scheduled against a deadline and two examined cases were received too late for inclusion in the analysis. Finally, there were 96 men whose only injury was at the level of the brachial plexus and 12 cases with Cs and C9 involvement, and since a hand analysis seemed necessary for these very complex cases, cards were not punched for them. There 11

OCR for page 1
These factors were organized in a classification which may be illustrated for the median nerve as follows: Injury-operation group number Set of cases 1 Complete suture, complete division, high 2 Complete suture, complete division* low 3 Complete suture, complete division* total 4 Complete suture, partial division, total 5 Complete suture, neuroma, high 6 Complete suture, neuroma* low 7 Complete suture, neuroma, total 9 Complete suture, all, high 10 Complete suture, all, low 11 Complete suture, all, total 12 Partial suture, all, total 15 Neurolysis, incomplete lesion, high 16 Neurolysis, incomplete lesion, low 17 Neurolysis, incomplete lesion, total Other nerves were classified in more or less detail, and, in all, 60 groupings were made covering all 7 major nerves, abbreviated as follows, M, U, R, P, T, SP, and ST. The general nature of the tabulation plan may be illustrated with reference to the motor chapter, which has the following parts: 1. For the representative sample only, each motor index was tabulated for each of the 60 injury-operation groups. 2. For the extended sample, each of 2 motor indices was run against each of the following characteristics of injury, again for each of the 60 injury-operation groups: Associated bone or joint injury, including type of healing. Associated arterial injury. Presence of chronic infection delaying repair. Presence of plastic repair at site of nerve lesion. Age. Specific site of lesion. 3. For the extended sample, each of 2 motor indices was run against each of the following details of management, again for each of the 60 injury-operation groups: Number of operations. Days from injury to definitive repair. Date of definitive repair. Hospital echelon of definitive repair. Tension on suture line reported by surgeon at definitive repair. Special operative features characteristic of definitive repair (e. g., bulb suture, transposition, etc.). Type of suture material, definitive suture. Type of neurolysis, definitive neurolysis. Type of cuff used at definitive operation. Use of stay suture at definitive suture. Length of surgical defect at definitive suture. Training of surgeon performing definitive operation. Motor response to electrical stimulation at definitive operation. Quality of nerve ends reported by surgeon at definitive repair. The above plan was embarked upon in systematic fashion in the expectation that a limited number of problems would be attacked later with more

OCR for page 1
attention to the interrelations among variables, once these first tables had been studied, and such proved to be the case. An early problem requiring solution even before the tables could be run was that of center variation. The magnitude of the tabulation task, and of associated computing, forbade any plan whereby all tables might be run separately for each center, and yet it was plain that certain observations might well have to be so handled, and there was no disposition to combine essentially unlike observations. Accordingly, once the punched cards had been made ready for use, the first task of the statistician was to investigate some of the more important observations from this standpoint. In the interests of homogeneity the analysis was confined to examined cases in which there was a single lesion of complete severance treated by complete suture. The follow-up observations chosen for study follow: Complaints as to pain and sensation. Complaints as to motor function. Complaints as to autonomic function. Pain threshold. Touch threshold. Position sense. Localization. Skin resistance and sweating. Functional status. Overall functional evaluation. Has occupation changed because of nerve injury? Handicap in present occupation. Need of patient for further treatment. Evidence for anatomic regeneration. Chronaxie. Tetanus ratio. Voluntary contraction. Strength of movement. Assessment of motor recovery. Although discrepancies among centers were observed, they were rarely of such a nature as to require, in the opinion of the statistician, a separate analysis of the material contributed by each center. The exceptions are as follows: 1. Complaints of all kinds (motor, sensory, and autonomic) were re- ported with quite variable frequency by center, and were analyzed by center. 2. Hyperpathia and dysesthesia, which were to be noted at the time of determining the pain threshold, did not have the same meanings for all centers, and were quite variably reported. Their analysis was done by center. 3. Chronaxie values of the several centers varied in a way suggestive of differences in calibration, and were studied by center. 4. Such marked variation was found in the observation of position sense and localization among the centers, and these modalities were studied in so few cases, that very little use of the data could be made. The center variation found in the British motor and sensory assessments 21

OCR for page 1
is discussed in connection with the presentation of appropriate data. There are, of course, two dangers associated with center variation. At the very least such variation will represent the introduction of an error which will be independent of other characteristics of the lesion and serve only to obscure relationships which might otherwise appear plain. Should center errors not be independent of other characteristics of the lesion, how- ever, or should one correlate two follow-up observations both affected by the errors of a particular center, then the door is opened wide to spurious relationships. Fortunately, the allocations are quite homogeneous from center to center, and the characteristics of injury and treatment were centrally coded, so that there is little room for the confounding of, say, a particular type of case with an erroneous follow-up observation in a par- ticular center. In the correlation of one follow-up observation with an- other, however, there is ample room for such confounding and we hope we have been sufficiently alert to it. The design of the punchcards also contributed greatly to the efficient handling of the data, especially in view of the multiplicity of observations on the various muscles; a single radial lesion, for example, might have dynamometer readings on 8 muscles. The key to simplicity proved to be the fact that the projected analysis would never require that the readings on one muscle be tabulated against those on another. This allowed the design of a single card carrying information about the injury, the definitive operation, and all the follow-up data except that only such follow-up data as pertained to a single muscle could be placed on a single card. Data of the latter type occupied columns 58-79 of the card, and any lesion would require 3 to 8 cards depending upon the number of muscles the examiners had agreed to include in their standard examination. If there were, as in the radial, 8 such muscles, then 8 muscle cards were prepared by re- producing into all of them the data of columns 1 to 57 in the basic design, and then adding in columns 58-79 of each the information for the particu- lar muscle which was in turn identified by the card (and muscle) identi- fication placed in column 80. Other intermediate and supplementary cards were required, but the bulk of the work was done on the muscle cards. Figures 3 and 4 contain the card number and show exactly how the material was placed on the final cards. Another methodological problem assigned to the statistician was to de- termine whether important bias had entered the material via incomplete follow-up. This problem was explored in three ways. First, on the basis of evidence that certain elements of injury and management were exercis- ing the role of determinants of end results, the examined and not examined cases were compared directly as to these elements. Second, on the assump- tion that the bias of nonresponse is progressively eliminated as one exerts increasing efforts to bring men under study, a comparison was made of men who reported for study without urging and of men who cooperated only after much persuasion. Third, the chairman reviewed and coded, on the basis of VA rating examinations, representative examined and 22

OCR for page 1
nonexamined cases of one center (New York) to permit comparison on key points. The complex nature of the sampling plan made it desirable to restrict the study of bias to the most homogeneous, representative cases. Accord- ingly, the tables on characteristics of injury and treatment were confined to lesions treated by complete suture, of men drawn from the Army Per- ipheral Nerve Registry, and of men resident within the sampling area defined for the representative sample. Nine characteristics of injury and treatment were chosen as the basis for comparisons of examined and non- examined cases, as follows: Identity of injured nerve. Site of lesion. Presence of chronic infection sufficient to delay definitive suture. Presence of associated arterial injury. Number of operations of definitive intent. Days from injury to definitive suture. Length of surgical gap at definitive operation. Use of cuff at definitive operation. Use of stay suture at definitive operation. The comparison as to identity of the injured nerve was done separately for high (elbow or knee and above) and for lower sites of injury, and in both instances only small, statistically insignificant differences were found. The study of site of lesion was confined to median, ulnar, peroneal, and tibial lesions and showed that follow-up was definitely better for the higher lesions, and especially so in the ulnar (table 4). Although the discrepancy seems quite clear, in that the combined probability is below .01 under the hypothesis of homogeneity of high and low lesions, it is not large in amount and has seemed of no great practical concern. Table 4.—Gross Site of Lesion and Percentage Examined at Follow-up, by Nerve, Completely Sutured Lesions in Representative Sample Percentage examined Nerve P High Low lesions lesions 86.9 86. 1 81.1 87.9 82.3 77.7 78.1 82.7 >.05 .02 >.05 >.05 Ulnar Peroneal Tibial All four combined at probability level OCR for page 1
by such infection and 17 for lesions without; this discrepancy has a prob- ability of about .02 under the null hypothesis. Associated arterial injury, number of operations, days from injury to operation, surgical gap, and use of cuff were all found to be unrelated to the chance of follow-up examina- tion. However, in the lower extremity, lesions for which the treatment included the use of a stay suture had a significantly higher follow-up percentage than lesions not so treated (90 v. 78 percent, which differ by an amount with a probability well below .01). Of the 9 characteristics employed in the comparisons on examined and nonexamined cases, only 3 (site of lesion, chronic infection, and use of stay suture) are associated with the chance of follow-up. To appreciate the import of these findings it is necessary to anticipate the results of the follow- up study by noting the relationship between the bias of follow-up and the apparent effect of each characteristic upon end result, as follows: Follow-up Performance percentage at follow-up Low lesions Low High. Lesions with chronic infection Low Low. Lesions with stay suture High Low. Even these three evidences of bias, that is, fail to suggest any general tendency for either the poorer or the better results to have been more frequently examined at follow-up. The second approach to follow-up bias made use of a classification of the effort required to bring the man in for examination. Three effort groups were contrasted: (1) men who came in response to a personal letter; (2) men who required more than one letter, but not as much as a personal visit by the Red Cross; and (3) men who required a personal visit by the Red Cross or intervention by the VA regional office. These three effort groups were then compared as to each of the following observations made at follow-up: Pain threshold. British summary of sensory regeneration. Skin resistance. Overall function of injured extremity. British summary of motor regeneration. In each instance, the comparisons were made by nerve, i. e., separately for median, ulnar, etc. In each instance except 1 the 3 effort groups differed by no more than chance. The exception was found in the peroneal nerve when overall function was made the basis of the comparison (table 5). However, no such relationship is seen in the other nerves, and in view of the large number of comparisons which were made on the five follow-up observations, to find one apparently striking discrepancy is not itself remarkable. Although the first two studies of follow-up bias provide an insufficient basis for denying the existence of such bias, they do suggest very strongly that follow-up bias is not an important problem in the study. 24

OCR for page 1
Table 5.—Percentage of Limbs With at Least 70 Percent of Overall Functional Capacity,1 by Extent of Follow-up Effort, Limbs With Complete Peroneal Sutures, Representative Sample Only Number of cases Percentage rated 70 percent or better as to overall func- tional recovery Effort Single letter ... 90 32 2 Two or more letters 31 51 6 Personal visit, or regional office intervention 17 94. 1 Total 138 44 2 1 The scale of overall functional capacity is described on pp. 351-353. The third study of follow-up bias rests upon a comparison of examined and nonexamined men from the standpoint of data found in VA claims folders. The New York allocation of complete lesions on the ulnar nerve, falling within the representative cross section and having no associated nerve lesions, was sampled at random to provide 31 nonexamined men and 26 examined. The VA claims folders contain all examinations made by VA examiners in connection with the adjudication of claims for compen- sation, and the chairman of the study group undertook to assess each case on the basis of two of the most summary scales employed in the follow-up study: (1) Dr. Lewey's scale of functional recovery (pp. 351-353); and (2) the British scale of motor recovery (p. 75). For the examined cases these supplementary ratings were accomplished in ignorance of the actual judg- ments reached by the New York group on the basis of its detailed follow- up examinations, and it was of more than passing interest to compare the two independent ratings. Since Dr. Lewey's scale is a quantitative one, it seemed natural to correlate the two ratings; the correlation coefficient is + .80. The British motor ratings agree even more closely; in 18 of the 26 cases the same rating was assigned and in 6 the ratings were but one step apart on the British scale. Although the samples are small, the agree- ment inspires confidence in the use of the ratings based on the VA claims folder to compare the examined and nonexamined cases. For both motor recovery and overall functional recovery the sample of examined men presents a significantly more favorable picture than the nonexamined. Table 6 presents the data on the British scale of motor re- covery; the two samples differ by an amount which has a probability below .01. Table 7 provides the parallel data on Dr. Lewey's scale of functional recovery; the two samples differ significantly (P = .04). Since the testi- mony of this experiment is clearly in favor of the notion that the follow-up sample is somewhat biased in the direction of more favorable results, it is 403930—87-

OCR for page 1
important to estimate at least roughly the magnitude of the error of esti- mation arising out of use of data on only those men who could be brought in for examination. This may be done easily, although the samples are too small to permit it to be done very reliably, by weighting the sample of examined men by 75 percent and the sample of nonexamined by 25 per- cent. This process leads to the estimates presented in table 8 for the British scale of motor recovery and in table 9 for Dr. Lewey's scale of over- all functional recovery. In each table the biased estimate is derived from the men who were examined, and the unbiased estimate was obtained from both samples by the weighting procedure described. Although the evi- dence of tables 6 and 7 is clearly to the effect that the examined men made the better recovery, the percentage of men actually brought under exami- nation is so high that the actual bias does not seem very large. Table 6.—Comparison of Examined and Unexamined Pure Ulnar Lesions as to British Classification of Motor Recovery Evaluated From VA Claims Folders, Representative Sutures From New York Allocation British motor classification Exam- ined Not ex- amined Total At most proximal muscles acting against gravity, no return of power in intrinsics Number of lesions Proximal muscles against gravity, and perceptible contrac- tion in intrinsics 4 20 8 24 21 Return of function in both proximal and distal muscles such that all important muscles act against resistance, or better 13 9 3 12 Total 26 31 57 Table 7.—Comparison of Examined and Unexamined Pure Ulnar Nerve Lesions as to Overall Functional Recovery Evaluated From VA Claims Folders, Repre- sentative Sutures From New York Allocation Overall functional recovery (percent) Exam- ined Not ex- amined Total Number of lesions 60 or less 8 17 10 4 25 17 15 70 7 11 80 or more ... Total 26 31 57

OCR for page 1
As a prelude to the coding experiment on the VA claims folders the VA numerical disability ratings in effect at the time of the follow-up study were obtained for fairly large samples of men in the New York allocation, both the examined and the unexamined. The two groups proved to be indistinguishable in their VA disability ratings. Table 8.—Comparison of Biased and Unbiased Estimates of Motor Recovery Following Ulnar Nerve Suture, Based on Study of VA Claims Folders Biased (examined men only) Motor recovery (British scale) Unbiased (all men) Percent Percent 3.7 No contraction . 0 Perceptible contraction, proximal only 0 4.6 15.0 21. 1 Proximal against gravity* perceptible contraction in intrinsics. . Proximal and distal against resistance 50.0 35.0 43. 1 27.5 As above, plus some synergic and isolated movements 0 0 Complete recovery 0 0 Total . .... 100.0 100.0 Number of men studied 26 57 Table 9.—Comparison of Biased and Unbiased Estimates of Overall Functional Recovery Following Ulnar Nerve Suture, Based on Study of VA Claims Folders Biased (examined men) Unbiased (all men) Functional recovery (percent) Percent 0 Percent 0 0 10 0 0 20 0 0 30 3.8 3.7 40 0 0 50 0 0 60 27.0 33.9 70 27.0 28.4 80 30.8 23.8 90 11.5 10.1 100 0 0 Total 100.1 99.9 Number of men studied 26 57 17

OCR for page 1
In summary, several separate studies have been made on the possible extent of any bias arising out of incomplete follow-up, and only one provides convincing evidence of bias. Moreover, when account is taken of the fact that 75 percent of the allocated men were examined, the suggestion is strong that any bias is not very extensive. Once the tables for a given chapter had been run, it fell to the statistician to analyze them from the point of view which dictated the original plan, and to prepare a written summary of both the tabular data and his analysis of them. Methods of analysis have been those generally used in testing hypotheses and need not be elaborated upon here; acknowledgment is made, however, to Mr. Seymour Jablon of the Follow-up Agency who served as mathematical consultant on numerous points in the analysis and effected great savings in the processing costs. Several statistical problems which must be confronted in the analysis deserve mention here also. In seeking to determine whether a factor, e. g., presence of associated bone injury, has some influence upon, say, motor recovery, one may approach the analysis via seven separate nerves, asking the same question about each, and being content to learn, for example, that there seems to be evidence of an effect in certain nerves but not in others. Or one may utilize the data on the different anatomic structures in seeking to learn if there be an effect of a general nature, applicable to all nerves. It is from the latter point of view that most of the analysis has been done, and exceptions are noted in the text. Another statistical problem concerns the use of data on several muscles innervated by the same nerve to enable conclusions to be reached about that nerve. The observations on one muscle are not inde- pendent of those on another in the same man, and the data provided by the several muscles cannot be combined on the basis of the assumption of independence without doing violence to the facts. Although alternative statistical techniques are helpful in this situation, a great deal of judgment is called for in the interpretation of the data, and this is one reason why, especially in the motor chapter, so much of the information is presented in the form in which it was obtained, i. e., by muscle. A third statistical problem, by no means unique to these data but es- pecially troublesome in evaluating the influence of factors associated with management, arises out of the association of one factor with another, e. g., time and surgical gap, so that apparent effects may disappear when analyzed in more detail or, what is worse, may actually be confounded to such an extent that their independent existence cannot be demonstrated by the data. One cannot, of course, be sanguine about an ex post facto study of treatment effects in routine clinical material, but must expect continually to be misled and mistaken by apparent relationships which are in fact based upon the factors governing selection of treatment in a clinical (not experimental) situation in which the physician undertakes to do what seems best for the particular patient. The interest in the details of treat- ment, however, will not allow their neglect even though the prospects for 28

OCR for page 1
definitive conclusions may be poor. One can hardly experiment with peripheral nerve lesions in the human. G. PREPARATION OF THE FINAL REPORT The plan developed at the Boston conference has been followed in its general outline. A first draft of each chapter has been prepared by the investigator responsible for the subject, and then reviewed and edited by the chairman with the assistance of the statistician. The statistician, in turn, has reviewed all the text from the statistical point of view and verified the statistical accuracy of such data and conclusions as are presented. Finally, the entire investigative group has reviewed the manuscript as a whole. H. SOME DEFICIENCIES SEEN IN RETROSPECT No cooperative study of such magnitude as the present one can be viewed in retrospect without regrets being felt over what now seem to be defici- encies, and to review a study from this standpoint may seem folly. How- ever, we have done so, in a limited way, not to persuade ourselves of the inevitability of what has been done but for whatever value it may have for other clinical investigators who must embark upon large enterprises where advance planning is of especial importance. Had the many important purposes of the study been made more explicit at the beginning, all subse- quent planning would have been easier and more effective, and the study a less expensive and time-consuming one to complete. Early pilot work on the content of the follow-up examination, with more recognition of the need for uniformity of concepts and examining proced- ures, would have permitted the drafting of the statistical code before the main series of examinations began and would have otherwise hastened and facilitated the successful conclusion of the study. More adequate advance planning would undoubtedly have enabled the investigators to approach their objectives on the basis of a smaller volume of observations per case, and to enter upon the final analysis in a leisurely and experimental fashion more in keeping with the complex interrelations among the variables under study. A major deficiency, in the opinion of all the investigators, was the necessity for each center to expend a major portion of its total resource on inducing and scheduling men to report for examination. The final score of 75 percent examined is a high one, but it was attained at great cost and delay in the entire undertaking; men drawing compensation from the Federal Government should be more accessible to government- sponsored investigation. In a cooperative clinical study of such magnitude, it would be desirable, after preparation of a preliminary scheme for examination and coding, for a pilot group of patients at each center to be studied under the personal direction of the permanent director who would be responsible for the train- ing of its professional and technical personnel. The pertinent results of the

OCR for page 1
pilot study should then be subjected to a searching analysis by the intended author of each chapter of the final report, who, after consultation with the statistician, would prepare the examination procedure and coding system to be adopted. Another improvement would be the provision of a full- time deputy, working closely with the chairman, who could make frequent visits of several days to each center, during which he would participate actively in the examination and coding of patients. The preparation and correlation of the final manuscripts would also be greatly aided by regular conferences between the traveling full-time deputy and each author. 30