Marsh 1WO DEFENSE HEALTH CARE Military Physicians’ Views on Military Medicine I 111111 140942 GAO/HRI)-90-l Human Resources Division B-238360 March 22, 1990 The Honorable Beverly B. Byron 1 Chairman, Subcommittee on Military Personnel and Compensation Committee on Armed Services House of Representatives Dear Madam Chairman: This report, prepared at your request, discusses the factors that most influence military physicians to leave the, military and reviews the extent to which military physicians left the service since 1986. In addition, the report illustrates how increasing physician compensation and reducing the time physicians spend performing nonmedical tasks could diminish the likelihood of attrition. We are sending copies of this report to the Secretary of Defense, the service secretaries, and interested congressional committees. We will also provide copies to the commanding officer at each military treatment facility and to other interested parties on request. This report was prepared under the direction of David P. Baine, Director, Federal Health Care Delivery Issues, who can be reached on (202) 275-6207 if you have any questions about the report, Other major contributors are listed in appendix IV. Sincerely yours, Lad uuAGtLk\ -& Lyy4h Lawrence H. Thompson Assistant Comptroller General , I Executive Summq The percentage of military physicians leaving the service increased in recent years, from 13.7 in 1986 to 16.6 in 1988. This has caused concern in the Congress and the military services because increases in the number of active-duty physicians leaving the military may affect the Department of Defense’s (DOD) ability to meet combat medical require- ments in the event of war and its ability to provide services to benefi- ciaries in peacetime. To assist in structuring a long-term solution to the military physician attrition problem, the Chairman, Subcommittee on Military Personnel and Compensation, House Committee on Armed Services, asked GAO to identify the reasons why military physicians are leaving the service in increasing numbers. GAO sent a questionnaire to about 1,500 active-duty physicians to assess l the likelihood that physicians will leave the military, and . the factors that most influence physicians’ decisions to leave the service. I DOD'S $13 billion per year health care system employs over 13,000 Bdckground active-duty physicians and serves about 9 million people. Physician retention has concerned the Congress and DOD for some time. To help reduce attrition, the Congress, in the 1989 National Defense Authorization Act, established a bonus for many physicians who agreed to remain in the service for 2 additional years. As required by the 1989 National Defense Authorization Act, DOD sub- mitted a report to the Congress that discussed issues affecting retention of military physicians and proposed a pay program to address the com- pensation gap with the civilian sector. DOD also submitted a follow-on report that examined noncompensation issues. Physicians’ intentions to leave the service over the next several years Results in Brief parallel DOD'S historical attrition rates. Almost one-half of the active- duty physicians indicated at least a 70-percent probability of leaving the service when they become eligible and about two-thirds voiced a 60- percent chance of leaving. Physicians reported dissatisfaction with many aspects of military medicine. At least one-half were dissatisfied with their compensation, the availability of administrative and health support personnel, hospital Page 2 GAO/HRDBO-1 Views on Military Medicine ISxecutive hmmary equipment, the ability to provide continuity of care to patients, having to perform excessive amounts of quality assurance tasks, and the oppor- tunities to attend professional meetings and training. Dissatisfaction with certain factors does not necessarily result in a phy- sician’s decision to leave the service. GAO’S analysis of several of these factors indicates that active-duty physicians’ intentions to leave are influenced by: (1) the time spent on nonphysician tasks, (2) a gap between military and civilian physician compensation, and (3) the lack of opportunity to practice in their primary specialties. The number of hours that initial obligees are required to devote to readiness training also affects their intentions to leave. Physicians who are beyond their initial obligations are influenced by the number of unwanted permanent changes of station. Further analysis suggests that, among these factors, the probability of physicians leaving military service can be most effec- tively reduced by increasing compensation, by decreasing the time phy- sicians spend on nonphysician tasks, or both. I GAO’s Analysis Many Physicians Will Likely Leave the Military Wheb Eligible to Do So Table li Phyalcians Planning to Leave the Military Upon Completion of Their Figures are percentages Obligation Stated probability of leaving Physician category 70% or greater 50% or greater Overall 47 62 Army 41 58 Navy 50 65 Air Force 52 63 Obstetricians/Gynecologists 60 74 Surgeons 51 67 lnternal medicine 46 64 Primarv care 42 56 Support medicine 71 78 Initial militarv obliclation 62 77 Beyond initial obligation 28 42 Page 3 GAO/HRD-99-l Views on Military Medicine Executive Summary Physicians Dissatisfied Slightly more than one-half of the physicians said they were dissatisfied With Several Aspects of with their pay and believed that private sector physicians earned 25 to 100 percent more than they do. Support medicine Physicians, such as Mi$itary Medicine anesthesiologists and radiologists, were the most dissatisfied and pri- I mar-y care doctors were the least dissatisfied with their pay. (See p. 22.) Table 2 shows the extent of physician dissatisfaction with other aspects of military medicine. Tab/e 2: Phyricians Dissatisfied With Ceyin Aspects of Military Medicine Percent of Aspect of military medicine physicians Too few clerks, receptionists, and secretaries 86 Too few nurses, corpsmen, and orderlies 75 Inadequate equipment in hospitals 65 / Unable to provide continuous care 62 Insufficient opportunities to attend professional meetings -63 Excessive quality assurance tasks 63 Too little emphasis on medical proficiency in promotion decisions 60 Factors Influencing Because questionnaire respondents identified compensation and the lack Physicians Intentions to of support personnel as primary inducements to leave, GAO'S analysis focused mainly on how these two factors affect physicians’ probability Leave of leaving. For example, increasing the average salary by $10,000 for physicians serving under their initial obligations reduces their average probability of leaving by an estimated 14 percentage points. The same salary increase for other obligees causes an estimated lg-percentage- point reduction in their stated probability of leaving. (See p. 28.) Initial obligees spend, on average, 11.3 hours per week on nonphysician tasks, such as nursing, secretarial, or administrative duties. Other obli- gees spend, on average, 8.4 hours per week on these tasks. Eliminating the time spent on nonphysician tasks decreases the stated probability of leaving by 5 percentage points for initial obligees and by about 10 per- centage points for other obligees. (See p. 30.) Other factors affect physicians’ stated probability of leaving, but their effects are small. For example, reducing the number of unwanted per- manent changes of station to 0 achieves a 3-percentage-point decrease in probability of leaving for other obligees. Eliminating readiness training entirely yields a l-percentage-point decrease in probability of leaving Page 4 GAO/HRD+O-1 Views on Military Medicine Executive Summary for initial obligees. Finally, if physicians’ dissatisfaction with their abil- ity to maintain specialty skills was eliminated, the probability of leaving would decrease, on average, by about 1 percentage point for initial obli- gees and 4 percentage points for other obligees. (See p. 32.) GAO is making no recommendations. Recdmmendations As requested by the Chairman, GAO did not obtain written agency com- Agency Comments ments on a draft of this report. However, the views of responsible DOD officials were sought and incorporated where appropriate. Page 5 GAO/HRDN-1 Views on Military Medicine @dents Ekecutive Summary Cl+apter 1 Introduction Military Physician Force Physician Attrition Objectives, Scope, and Methodology Chapter 2 14 Characteristics of Medical Education and Experience Obligation Status 14 16 Military Physicians How and Why Physicians Entered the Military 17 1 Physicians’ Primary Activities 18 Opportunity to Practice Medical Specialty 19 Chapter 3 20 Ektent of Physician Many Physicians Indicated a High Probability of Leaving the Military 20 Dissatisfaction With Physicians Dissatisfied With Their Compensation 21 Military Medicine Inadequacy of Support Staff 22 Physicians’ Views on Other Aspects of Military Medicine 24 Cl+apter 4 26 Fqctors That Most Impact of Pay on Stay/Leave Decision Impact of Nonphysician Tasks on Stay/Leave Decision 27 29 Influence Physicians’ Ability to Maintain Proficiency in Area of Specialization 31 Plans About Military Combat Readiness Training 31 Permanent Changes of Station 31 Service Summary 32 Appendixes Appendix I: Physicians Planning to Leave the Military 34 Upon Completion of Their Obligation Appendix II: Physician Views on the Adequacy of Health 35 and Administrative Support Personnel Appendix III: Survey and Data Analysis Methodology 38 Appendix IV: Major Contributors to This Report 50 Tables y Table 1: Physicians Planning to Leave the Military Upon 3 Completion of Their Obligation Page 6 GAO/HRD-!%I Views on Military Medicine Contents Table 2: Physicians Dissatisfied With Certain Aspects of 4 Military Medicine Table 1.1: Physician Losses by Service 11 Table 1.2: Ph sician Attrition Rates by Service 12 Table 2.1: Yea6 of Experience of Military Physicians, by 16 Service Table 2.2: Physicians Saying They Spent Time Performing 19 Certain Tasks Table 3,l:Physicians Dissatisfied With Pay, by Medical 22 Specialty Group Table 3.2: Physicians Who Believe Health Staff Is 23 Insufficient Table 3.3: Physicians Who Believe Administrative 23 Support Personnel Is Insufficient Table 4.1: Estimated Effect of a $10,000 Salary Increase 28 on Initial and Other Obligees’ Stated Probability of Leaving the Military Table 4.2: Estimated Effect of a $10,000 Salary Increase 28 on Various Specialists’ Probability of Leaving the Military (Initial Obligees) Table 4.3: Estimated Effect of a $10,000 Salary Increase 29 on Various Specialists’ Probability of Leaving the Military (Other Obligees) Table 4.4: Estimated Impact of Eliminating Nonphysician 30 Tasks on Initial and Other Obligees Table 4.5: Estimated Impact of Eliminating Nonphysician 30 Tasks on the Probability of Leaving the Military, by Medical Specialty Group Table 4.6: Impact of Potential Personnel Policies on 32 Retention of Military Physicians Table I. 1: Physicians Planning to Leave the Military Upon 34 Completion of Their Obligation (By Service) Table 1.2: Physicians Planning to Leave the Military Upon 34 Completion of Their Obligation (By Medical Specialty Group) Table 1.3: Physicians Planning to Leave the Service Upon 34 Completion of Their Obligation (By Type of Obligation) Table II. 1: Internal Medicine Physicians’ Views on 35 Adequacy of Support Staff Table II.2 Obstetrician/Gynecologists’ Views on Adequacy 36 of Support Staff Page 7 GAO/HRD-90-l Views on Military Medicine Contents Table 11.3:Primary Care Specialists’ Views on Adequacy 36 of Support Staff Table 11.4:Medical Support Specialists’ Views on 37 Adequacy of Support Staff Table 11.5:General and Other Surgical Specialists’ Views 37 on Adequacy of Support Staff Table III. 1: Random Sample Selection-By Service and 39 Stratum Table 111.2:Estimates of Regression Equations for Initial 46 and Other Obligees Figures Figure 2.1: Percent of Physicians Who Completed 15 Residency and 1 Year of Post-Graduate Training Figure 2.2: Percent Distribution of Physicians by Type of 16 Obligation Figure 2.3: Physicians’ Obligation Status by Service 17 Figure 2.4: Five Most Important Reasons for Physicians 18 Entering the Military Figure 3.1: Physicians’ Interest in Civilian Employment 21 Abbreviations AFHPSP Armed Forces Health Professions Scholarship Program CHAMPUS Civilian Health and Medical Program of the Uniformed Services DOD Department of Defense GAO General Accounting Office GME graduate medical education Ob/Gyn obstetrics/gynecology USUHS Uniformed Services University of the Health Sciences Page 8 GAO/HRBI)O-1 Views on Military Medicine Page 9 GAO/HRLNO-1 Views on Military Medicine The Department of Defense (DOD) health care system has two primary objectives. The first is to maintain the health of the active-duty force of the uniformed services and be prepared to attend the sick and wounded in wartime.1 The second is to provide medical care to eligible dependents and retirees when space, facilities, and staff are available. DOD maintains over 500 military treatment facilities, ranging in size from small clinics with limited capabilities to large hospitals with extensive capabilities and medical teaching programs. Military treatment facilities serve almost 9 million beneficiaries: 2.3 million are active-duty mem- bers, 2.8 million are dependents of active-duty members, and 3.7 million are retirees and dependents of retired or deceased members. DOD records show that the estimated cost of medical care at military treatment facilities, exclusive of medical facility construction costs, was about $10.2 billion for fiscal year 1989. When care is unavailable at mili- tary treatment facilities, nonactive-duty beneficiaries may seek care from civilian providers through the Civilian Health and Medical Pro- gram of the Uniformed Services (CHAMPUS). CHAMPUS costs in fiscal year 1989 were about $2.5 billion. To help meet its health care responsibilities, at the end of fiscal year Military Physician 1988 DOD had approximately 13,000 active-duty physicians: 5,295 in the Force Army, 3,983 in the Air Force, and 3,929 in the Navy. Each service has a medical department headed by a Surgeon General who is responsible for the activities of the services’ medical facilities. The Assistant Secretary of Defense (Health Affairs) is responsible for providing overall supervi- sion and policy guidance for DOD medical care activities. There are three major ways in which the military services recruit mili- tary physicians: (1) the Armed Forces Health Professions Scholarship Program (AFHPSP), (2) the Uniformed Services University of Health Sci- ences (IJSUHS), and (3) direct recruitment of trained physicians. For the first two sources, the military services finance the physicians’ medical education in exchange for their agreement to practice medicine as an active-duty military physician. In general, the number of years of active duty required matches the number of years of medical training financed by the military for AFHPSP physicians. The minimum obligation is 2 ‘The uniformed services include the Army, Navy, Air Force, Marine Corps, Coast Guard, and Com- missioned Corps of the Public Health Service and of the National Oceanic and Atmospheric Administration. Page 10 GAO/HRD-90-l Views on Military Medicine Chapter 1 Introduction years. For USUHS graduates, the active-duty obligation is 21 months for each year, or portion thereof, spent in medical training, except that in no case is the minimum obligation less than 27 months. Therefore, physi- cians who joined the military either through AFHPSP or USUHS are in obli- gated status for several years after completing medical school. Generally, physicians’ pay-back periods do not start until completion of their post-graduate training, (e.g., residency). Volunteers (those who joined the military as fully trained physicians) do not have an obligated pay-back period. As table 1.1 shows, between fiscal years 1985 and 1988, the number of Physician Attrition active-duty military physicians leaving the services increased by 222 or about 12 percent.2 As of fiscal year 1988, DOD had 13,207 physicians on active duty. Table i.1: Physician Losses by Servlce Fiscal year Chan e 1985 1986 1987 1988 1985-19 8,8 Air Force 438 472 472 529 91 Army 404 518 552 511 107 Navy 431 426 448 455 24 Total 1,273 1,416 1,472 1,495 222 Note: Includes personnel released/retired during the reporting period and physicians on inter-service and intra-service transfers, but excludes physicians in GME As a proportion of the total number of active-duty physicians, attrition rose from 13.7 percent in fiscal year 1985 to 15.6 percent in fiscal year 1988, excluding physicians in GME programs. Attrition rates increased since 1985 in all three services, but the increases were greatest in the Army and Air Force as shown in table 1.2. The Army’s attrition rate rose by almost 25 percent and the Air Force attrition rate by about 13 percent. By contrast, the Navy’s attrition rate rose by about 5 percent. Overall, however, the Army’s attrition rates have consistently been the lowest among the services. 2These data do not include physicians who were in graduate medical education (GME) programs. GMETsometimes &led residency training, refers to the period physicians spend in post-graduate trammg in a medic&l specialty. Page 11 GAO/HRD-90-l Views on Military Medicine Chapter 1 Introduction / 1.2: Physician Attrition Rates by (Excluding GMEs) Fiscal Year 1985 1986 1987 1988 Air -~ Force 15.1 15.5 15.1 17.0 1.9 12.6 Army 11.3 13.9 15.0 14.1 2.8 24.8 Navy 15.2 14.7 15.4 16.0 0.8 5.3 Total 13.7 14.7 15.1 15.6 1.9 13.9 Note: These attrition rates reflect the year-to-year fluctuations caused by new recruitment of physicians as well as by physician losses. They exclude physicians in GME programs. To encourage more physicians to remain in the military, the National Defense Authorization Act for Fiscal Year 1989 established a medical officer retention bonus. Up to $20,000 per year is paid to physicians with 8 or more years of military service who agree to remain on active duty for at least 2 more years. The bonus program has been referred to by the Chairman of the Subcommittee on Military Personnel and Com- pensation, House Armed Services Committee, as a stop-gap measure to retain experienced military physicians. In addition, as the act directed, the Secretary of Defense reported on the adequacy of the existing compensation package for health care profes- sionals and submitted a proposal for a compensation system linked to earnings of physicians in the private sector. DOD also reported on noncompensation factors affecting physician retention. On August 19,1988, the Chairman, Subcommittee on Military Personnel Objectives, Scope,and and Compensation, House Committee on Armed Services, asked us to Meithodology develop and administer a questionnaire to physicians on active duty to elicit their views concerning factors that most directly influence their decision to remain in or to leave the military. The questionnaire we developed elicited information on several issues, including compensa- tion, support staff, time spent on various medical and nonmedical tasks, patient mix, training, professional development opportunities, nonmedi- cal aspects of military life, reasons for entering the military, and the probability of leaving it. We sent the questionnaire to 1,515 active-duty physicians, about 500 from each branch of service, based on a stratified random sample of all active-duty physicians except those in GME programs. We excluded phy- sicians enrolled in GME because they are several years away from mak- ing decisions about their plans to leave or remain in the military. In Page 12 GAO/HRD-90-l Views on Military Medicine chapter 1 Introduction consultation with DOD officials and GAO’S Chief Medical Advisor we grouped physicians into five strata: (1) obstetrics/gynecology (Ob/Gyn), (2) primary care (e.g., family practice, general medicine), (3) internal medicine, (4) surgical specialties, and (5) support medicine (e.g., anes- thesiologists, radiologists). Appendix III provides additional examples of medical specialties included in each strata. The survey was conducted between November 1988 and January 1989 and achieved a response rate of 85 percent.” The respondents are repre- sentative of military physicians in general and, therefore, their responses can be generalized to all active-duty military physicians. In order to gain further insight into the factors that influence physi- cians’ probability of leaving the military, we analyzed survey responses using multiple regression analysis. This analysis focused on the correla- tion between physicians’ stated probability of leaving and several inde- pendent variables, such as compensation and hours spent on nonphysician tasks. This method also allowed us to predict the change in the probability of leaving that would be associated with a given change in an independent variable, assuming the other factors that might affect probability of leaving were held constant and physicians do what they say they will. Appendix III describes in more detail the sample selection and data anal- ysis methodologies we used. Our work was performed between August 1988 and September 1989 in accordance with generally accepted gov- ernment auditing standards. “On March 16, 1989, GAO testified before the Subcommittee on Military Personnel and Compensa- tion, House Committee on Armed Services, on preliminary results of its survey. Page 13 GAO/HRD-90-l Views on Military Medicine I Chapter 2 Characteristicsof Military Physicians This chapter describes the active-duty physician population and how they spend their time as depicted by the questionnaire respondents. Besides demographic data, information is presented on physicians’ medi- cal education and experience, the proportion of physicians in each medi- cal specialty group, how and why physicians entered the military, and the extent to which they practice in their specialty. Overall, military physicians are relatively young-the median age is 35 years, Eighty-eight percent are male. Nearly all received medical train- ing in the United States and most are certified by medical specialty boards in one or more specialties. The majority of physicians are in pri- mary care specialties. Generally, military physicians reported that they practiced in their specialties or subspecialties and that the primary activity through their career has been to deliver medical care or to engage in clinical activities. Mebical Education and United States; 7 percent were graduates of foreign medical schools. Fig- Experience ure 2.1 shows the level of professional medical education achieved by physicians in each of the services. Page 14 GAO/~99-l Views on Military Medicine .. Chapter 2 Characteristics of Military Physicians Flgure 7.1: Percent of Physlclans Who Compl?ted Residency and 1 Year of Percent of Physiclam, PostGraduate Training I Ait Amy Navy Total Force 1 1 post-grad Year 1 one full residency About 60 percent of all physicians are certified by medical specialty boards in one or more specialties. Across all services, the largest number of board-certified physicians is in family practice (primary care stra- turn)-96 percent of all family practitioners are board certified. Emer- gency medicine physicians (primary care stratum) have the lowest proportion of board certification-27 percent are board certified in their specialty. On average, physicians in the military have practiced medicine for about 10 years, including post-graduate training. Military physicians, in gen- eral, have practiced medicine in a nonmilitary setting for an average of 4 years, including physicians entering through AFHPSP who participated in internship and residency programs at civilian medical schools as well as those who joined the service as fully trained physicians. As shown in table 2.1, the total number of years military physicians have practiced medicine varies slightly among the services. Page 15 GAO/HRDQO-1 Views on Military Medicine I I Chapter 2 characteristics of Miku-y Physicians Table i2.1: Years of Experience of Military Physi :ians, by Service Figures are percentages Year of experience Branch of Less than 5 over 21 service -- years 6-10 years 11-15 years 16-20 years years Army 32 35 18 14 3 Navy 33 34 16 12 6 Air Force 41 23 22 12 3 Note: Columns do not add to 100 due to rounding Figure 2.2, shows the percent of military physicians serving their initial Obligation Status obligations; that is, “paying back” the military for financing their medi- cal education; the percent of military physicians serving under some other type of obligation incurred as a result of participation in physician special pay programs, additional training, or promotions; and the per- cent of military physicians serving under no obligation. FIgur? 2.2: Percent Distribution of Physicians by Type of Obligation No Obligation Initial Obligation Other Obligation Physicians under initial obligation are those paying back for their medical education with a comparable number of years of military service. Figure 2.3 shows the obligation status of physicians by service affiliation. Page 16 GAO/HRD-90-l Views on Military Medicine . chapter 2 characterietlcs OiMllltary Physlclans FlgurG 2.3: Phy@lclane’ Obligation Status by S@vlco 90 80 70 60 so 40 20 20 10 0 Air AMY Navy Force L-l No Obligation Other Obligation Initial Obligation Responses to our questionnaire indicated that about 52 percent of all Ho% and Why active-duty physicians entered the military through AFHPSP. About 16 Ph$sicians Entered percent entered voluntarily as fully trained physicians. About 6 percent the’ Military came in through USUHS. The remaining 28 percent of physicians entered as draftees, through other draft-related programs, or joined the military in another capacity and later opted to pursue a military medical career. Figure 2.4 shows the major reasons why physicians entered military service. One-half of the physicians on active duty were uncertain about whether to pursue a career in the military when they entered. About 29 percent of them said that they had not planned to pursue a career in the military while about 21 percent intended to do so at the time they entered. Page 17 GAO/HRD-!+@l Views on Military Medicine chapter 2 Chruacteristica of Military Physicituw Flgurd 2.4: Five Meet Important Rearone for Phislcianr Entering the Military 78 Percent of Physicians 70 66 00 65 - 50 45 40 35 30 25 20 16 10 6 0 ‘1 Five most important mesons for entering / For almost three-fourths of physicians (71 percent), the primary activ- Physicians’ Primary ity throughout their career has been to provide medical care or to Activities engage in clinical activities. Of this 71 percent, almost three-fourths responded that providing medical care was their primary area of responsibility. About 17 percent said that management or administra- tion was their primary responsibility, and 12 percent were primarily involved in teaching, research, or other activities. There were virtually no differences among the three services in terms of physicians’ primary activities, except fewer Army physicians were involved in management or administrative activities than Air Force and Navy physicians. Table 2.2 summarizes how responding physicians said they spent their time during a typical work week. Page 18 GAO/HRD-So-1 Views on Miitmy Medicine Chapter 2 Charactdstice of MilItary Physicians Table 51.2:Physicians Saying They Spent Time @formIng Certain Tasks Figures are percentages Hours Der week Over I lark 0 l-5 6-10 11-20 21-40 40 Medical care/teaching 3 5 5 11 46 ___--- 30 Management/administrative 10 34 24 18 10 4 Research 77 17 4 1 1 0 Quality assurance tasks 17 65 15 3 0 0 Nonphysician health care - 33 44 16 6 1 0 Nonphysician administrative 16 53 24 6 1 0 Note: Nonphysician health care tasks are those normally performed by a nurse, orderly, or corpsman. Nonphysician administrative tasks are those normally performed by clerks, receptionists, secretaries, or administrative personnel. About 83 percent of physicians practiced in their primary specialty or Opbortunity to subspecialty to a great extent. About 14 percent practiced in their spe- Practice Medical cialty to some extent, and very few (about 3 percent) practiced in their specialty to little or no extent. The extent to which physicians practiced Sp@alty their primary specialty or subspecialty was similar in the Army, Navy, and Air Force. w Page 19 GAO/HRD-90-l Views on Military Medicine Chapter 3 E$tent of Physician DissatisfactionWith Military Medicine Military physicians indicated a substantial degree of dissatisfaction with various aspects of military medicine. Almost one-half of the respondents indicated that there was a 70-percent or higher probability that they will leave the military when they become eligible. Physicians expressed dissatisfaction with their compensation, the lack of administrative and health support personnel, the inadequacy of military medical facility equipment, the inability to provide continuity of care to patients, and too little emphasis being placed on medical proficiency in decisions to promote military physicians. The lack of support for professional train- ing and travel as well as requirements to perform some quality assur- ance tasks, which in the physicians’ view do not improve quality of care, are also sources of dissatisfaction. The opportunity to practice their medical specialty was an aspect of mil- itary medicine that physicians cited as an inducement to stay in the ser- vice. They also expressed general satisfaction with the control they had over choosing their duty stations. Forty-seven percent of the physicians reported that the probability of Many Physicians their leaving the military when their current obligation expires was 70 Indicated a High percent or greater. Sixty-two percent indicated at least a 50-percent Probability of Leaving probability of leaving. An estimated 3,000 of the physicians under obli- gation who report a 70-percent or higher probability of leaving will be the Military eligible to leave before the end of 1992. Attrition at the rate suggested by these numbers would correspond with DOD’S actual attrition rates over the last 3 years. Physicians serving under their initial military obligation reported about twice the probability of leaving as did physicians serving under other types of obligation. A higher percentage of obstetricians/gynecologists and support medicine physicians reported a higher probability of leav- ing the service than the other medical specialties. Army doctors indi- cated the lowest probability of leaving. Appendix I shows the stated probabilities of leaving by branch of service, medical specialty group, and type of obligation. Although we do not know whether physicians will act on their stated intention to leave the military, figure 3.1 shows that many of the respondents to our questionnaire have been contacted about civilian employment and about one-third have taken action toward obtaining civilian employment. Page 20 GAO/HRD-So-1 Views on Military Medicine Chapter 3 &tent of Physician Dissatisfaction With Military Medicine Figurd 3.1: Physicians’ Interest In Civlllan Emplo/yment 100 Pwcenl of Physicians 90 - 80 70 60 50 40 30 20 10 0 1 l-n L 2 L 3 4 1) Physician has been contacted about civilian position. 2) Physician has initiated contact about civilian position 3) Physician has applied or interviewed for a civilian position. 4) Other activities to pursue civilian position. Fifty-three percent of the physicians said that they were somewhat or Physicians very dissatisfied with their pay and 88 percent said that physicians in Dissatisfied With the private sector were better off than they were in terms of earnings. Their Compensation More specifically, a majority of the military physicians we surveyed believed that physicians in the private sector earned substantially more than they did. Dissatisfaction with compensation was similar among Army, Air Force, and Navy physicians. However, physicians’ views on pay and on civilian earnings varied considerably by physician specialty. As shown in table 3.1, support medicine physicians were the most dissatisfied with their pay and also were the most likely to believe that their civilian counter- parts earned significantly more. Page 21 GAO/HI&D-%1 Views on Military Medicine Chapter 3 Extent of Physician Dissatisfaction With Military Medicine Table/3.1:Physlclans Dlssatlsfled With Pay, by Medical Specialty Group I Sur ical Internal support All View/Issue Ob/Gyn_ specla3 ties medicine Primary care medicine physicians So&&hat to.very dissatisfied with pay 65 -- 57 66 - 46 74 53 .. -.-~ probability -.--.----.---~~ of leaving pay and 70-percent or higher Dissatiisfied t with 41 27 29 21 52 26 &iievi d they &rkd 25 _..... - _.- less- --.-. percent than their 0 .___ 1 2 9 0 5 civil ean counterparts ....--__--___~~~~_,____~~ .-_____ bel&k?d they earned between 25-100 percent 39 34 50 71 17 56 less; than their civilian counterparts Believed they earned between 100-300 percent 61 65 48 20 83 38 lessi than their civillan counterparts We asked physicians whether a change in their pay that indexed their earnings to comparable specialists in private practice would affect the probability that they would leave military medicine. Thirty-four percent said that a pay package indexed to earnings of physicians in private practice would have little or no effect on their decisions to stay in the military. But 61 percent said that their probability of leaving would be somewhat or greatly decreased if pay were indexed to that of physicians in private practice. With regard to whether a bonus would serve as an inducement to remain in the military, 71 percent of the physicians said that it would provide at least some inducement to remain. Nearly all military physicians were dissatisfied with the number of Inaldequacy of Support health or administrative personnel available to support them. The Staff responses to our questionnaire regarding inadequacy of support resources closely parallel our February 1989 report on the extent to which military physicians perform administrative and clerical tasks.’ Generally, physicians indicated that the number of health personnel staff, such as general duty nurses, physician assistants, X-ray and labo- ratory technicians, and corpsmen was insufficient to support their work. Table 3.2 shows the percentage of physicians who indicated either mod- erate or extreme dissatisfaction with current levels of health support personnel. ‘DOD Health Care: Extent to Which Military Physicians Perform Administrative Tasks (GAO/ - 9- 53, Feb. 13,1989). Page 22 GAO/HRD-m-1 Views on Military Medicine Chapter 3 Extent of Physician Dissatisfaction With MUhry Medicine Table 3.2: Physicians Who Believe Health Staff lai InsufficIent Type of 8taff Percentage of physicians Suraical nurses 37 Nurse anesthetists 27 Nurse practitioners 41 Nurse midwives --- 16 General dutv nurses 74 Physician assistants 45 X-ray technicians 47 Laboratory technicians 58 Corpsmen/orderlies 76 As shown in table 3.3, physicians also expressed concern about insuffi- cient administrative and clerical support. Table $3: Physicians Who Believe AdmlnJdtrative Support Peraonnel Is Type of personnel Percentage of physicians lnsuffieient Administrative officers - 30 Medical records clerks 65 Secretaries, clerks, receptionists 86 Anecdotes provided by our respondents illustrate the problem of lack of secretarial/typing support. One physician wrote: “Give me a receptionist/secretary to help out in the office. (I have to do -all my typ- ing. I have no typing support.)” Another physician stated: “I type all my own correspondence, call all my own patients, and do much more secretarial work than is necessary.” And another: “Finally a comment on the lack of secretaries, etc. No problem, thanks to the Army I’ve learned to type close to 80 words/min. and I’m expert at using the Mackintosh System. I’ve purchased my own system ($3,000) and my troubles are solved.” There are no significant differences of opinion among Army, Navy, and Air Force physicians with regard to the adequacy of the current levels w of administrative support personnel. When segregated by specialty groups, physicians viewed shortages of support personnel differently. Because different medical specialists require different types of health Page 23 GAO/HRD-90-l Views on Military Medicine Chapter 3 Extent of Physician Dissatisfaction With Military Medicine / support staff, variations exist among the types of staff they consider to I be in short supply. Appendix II shows these variations by specialty I groups. I About 66 percent of the physicians surveyed indicated dissatisfaction sicians’ Views on with the way military treatment facilities are equipped; 62 percent were r Aspects of dissatisfied with being unable to provide continuity of care to their Military Medicine patients; and about 60 percent were dissatisfied because they felt that too little importance is placed on medical proficiency in promotion I decisions. Over one-half of all military physicians indicated that they were some- what to very dissatisfied with quality assurance requirements. About 63 percent considered they spent more time than needed performing tasks associated with these requirements and about three-fourths also believed that these tasks did not improve the quality of patient care. Seventy-two percent of all military physicians believed that civilian quality assurance requirements are less stringent than those in the mili- tary. Very few (about 7 percent) believed the requirements were more demanding in the private sector and about 22 percent thought that requirements were comparable in both sectors. Another aspect of military medicine with which physicians reported dis- satisfaction was the inability to attend as many professional conferences and seminars as their colleagues in the private sector. About 63 percent of all physicians are somewhat to very dissatisfied with the amount of time and travel funds for professional activities the military provides, Although 89 percent reported receiving time off or travel funds for this purpose at least once in their career, the average number of professional events physicians attend each year is slightly over one. About 70 per- cent believed that the military’s support for their professional develop- ment was less than adequate. When asked to identify the most important inducements to remain in the military, the most frequently cited reason was the opportunity military physicians have to practice their field of medicine/surgery or, if trained in more than one field, the opportunity to practice the field of medicine they prefer to practice. About 50 percent of the physicians said that the opportunity to pursue a medical specialty was an inducement to remain in the military. Almost 60 percent said that ability to maintain profi- ciency was also an inducement to stay. Other inducements to stay in the Page 24 GAO/HRD-90-l Views on Military Medicine Chapter 3 Extent of Physician Dissatisfaction With Military Medicine military include a desire to help assure that members of the Armed Ser- vices receive quality medical care and the fact that military physicians do not have to concern themselves with the patients’ ability to pay for care. On the whole, physicians are satisfied with the control they have over choosing the locations to which they are assigned. Only 11 percent of all physicians indicate that they are dissatisfied with the location of their duty station. Most physicians (about 84 percent) had undergone a permanent change of station. On average, physicians changed stations about three times during their military career but were generally willing to move. Forty- three percent of all physicians had no preference about remaining in their previous assignment or moving. Deployments-being stationed for an extended period of time in an operational setting-occurred with about one-third of all physicians. The remaining two-thirds had never been deployed. The average dura- tion of deployments was related to service affiliation. The average dura- tion of Army and Navy deployments was about 5 months compared with 3 months for the Air Force. Approximately 75-percent of all Army and Air Force physicians had never been deployed compared with 41 per- cent of Navy physicians. In general, physicians do not express dissatis- faction with the frequency of deployments in any of the services. Page 26 GAO/HRD-90-l Views on Military Medicine Chapter 4 Fa$ms That Most Influence Physicians’ Plans Ak$outMilitary Service Although many considerations are likely to affect an individual physi- cian’s decision to leave the military, some are more important than others. To identify the relative importance of various factors, we used multiple regression to analyze the responses of the physicians we sur- veyed. In this way, we estimated the change in the stated probability of leaving associated with a given change in each of the independent fac- tors we c0nsidered.l Our analysis identified several factors that have a statistically significant impact on a physician’s stated probability of leaving. Lower probabilities are associated with l higher levels of military pay (relative to civilian compensation), . less time spent on nonphysician tasks, . the ability to maintain proficiency in a medical specialty, l less time spent on combat readiness training, and . fewer undesired permanent changes of station. The analysis was performed separately for physicians serving under their initial obligation and for other obligees.2 The two groups were ana- lyzed separately because they are likely to differ in their taste for mili- tary life and, consequently, this may affect not only the probability of leaving the military, but also the individuals’ responsiveness to factors such as military pay or combat readiness training. On average, physi- cians who serve beyond their initial obligations may be expected to have a greater predilection for military service than physicians still serving ’ In its 1988 Pay Study, DOD reported that increases in military pay are associated with measurable reductions in attrition rates, although DOD’s estimate of the magnitude of the pay effect is smaller than our estimate. In general, DOD’s approach, both in terms of methodology and variables consid- ered, is similar, but not identical, to the approach used in this report. One important difference between the two studies is that DOD’s data contained information on actual attrition rates, whereas our study measures the impact of various factors on physicians’ stated intentions to leave the military. “At 6 to 6 years of service, most military physicians face the first stay/leave decision. This point marks the end of the “pay back” period required by the agreement to serve as a miliary physician for a number of years in exchange for having received a medical education from the military. In this report, all grade O-4 physicians who responded “yes” to the question “Are you currently serving under your initial obligation” are included in the initial obligees category. Those physicians who do not leave the military at the end of their initial obligation incur additional obligations by virtue of promotions, eligibility for certain special pays, permanent changes of station or additional education. The length of this additional commitment varies. For this report, we included in the other obliges category all grade O-6 and O-6 physicians and all those grade O-4 physicians who responded “yes” to the question “Are you serving under any other type of obligation.” About 10 percent of the survey participants responded that they are under no obligation, including some who at the time of our survey were eligible for retirement. These physicians are included in the other obligees category. Physicians under no obligation were not analyzed separately because the sample size is too small to provide statistically meaningful results. Page 26 GAO/HRD-90-l Views on Military Medicine Chapter 4 Factma That Most Influence Physlciane’ Plaua About Military Service under their initial obligation -many of whom will decide to leave the military at the end of that obligation. Three factors were statistically significant influences for both groups of physicians: military pay, the number of nonphysician hours, and ability to maintain proficiency. For initial obligees the stated probability of leaving is also affected by the number of hours per month spent on read- iness training. Among other obligees, the intentions to leave are affected by the number of undesired changes of duty station. Of all the factors considered, military pay and nonphysician hours appear to be the most important in achieving meaningful reductions in physicians’ intentions to leave the military. While other factors are sta- tistically important, even the complete elimination of the difficulty in maintaining proficiency, unwanted changes of station, and hours per month spent on readiness training result in minor reductions in the probability of leaving. Physicians’ responses to questions about the earnings of civilian physi- Impact of Pay on cians point to the importance of pay for the stay/leave decision. In gen- Stay/Leave Decision eral, military physicians stated a higher probability of leaving if they thought that their civilian counterparts were much better off finan- cially. For example, 83 percent of the medical support specialists believed their civilian counterparts earned at least 100 percent more than they did and 71 percent indicated a high probability (70 percent or more) of leaving the military when eligible to do so. Conversely, only 20 percent of primary care specialists believed their civilian counterparts earned at least 100 percent more than they did and only 42 percent indi- cated a 70-percent or higher probability of leaving when eligible. The regression analysis shows that when civilian compensation and other factors are held constant, both initial and other obligees’ stated probability of leaving decreases as military pay increases. This inverse relationship between military pay and intentions to leave is statistically significant. (Statistical significance is discussed and regression coeffi- cients are shown in appendix III.) Estimated Effect of a Table 4.1 shows the estimated impact of a $10,000 increase in pay on $10,000 Salar3;Increase the stated probability of leaving for initial and other obligees earning the mean salary. Page 27 GAO/HRD-90-l Views on Military Medicine I,/ Chapter 4 Factora That Most Influence Physicians’ Plans About Mllltaq Service Table 4.1: Estimated Effect of a $10,000 Salary Increase on Initial and Other Estimated percent Percentage- Obligees’ Stated Probability of Leaving probability of leavin point decrease the Mqlitary Current percent after $lO,OO8 in probability of Average military pay probability of leaving increase leaving $73,259 (initial) 80.5 74.2 14.3 $84,257 (Other) 52.7 33.9 18.8 I Note: The source for salary data in this table is DOD’s Report to Congress entitled: Health Professionals I Special Pays Study, December 1, 1988. Salary figures are based on mean initial and other obligees salanes usrng I-rscal Year 1989 regular military compensation plus variable special pay, additional spe- cial pay, board-certified pay, and incentive special pay for eligible specialties. By increasing the compensation of initial obligees $10,000, a decrease in the probability of leaving in the range of 11 .Oto 17.7 percentage points can be achieved, depending on individual salary levels. Similarly, a 7.1- to 20.3-percentage point reduction can be obtained for other obligees. If physicians act in accordance with their stated intentions, at current reported salary levels we expect that approximately 877 out of 1,000 physicians under initial obligation would leave the military while 123 physicians would stay. After a $10,000 pay increase, we expect that approximately 735 would leave and 265 physicians would stay. This represents an expected gain of 142 physicians. For other obligees, we expect that 538 of 1,000 physicians would leave the military while 462 would stay. After a $10,000 increase, we expect that 375 would leave while 625 would stay. This represents an expected gain of 163 physicians. Est$matedEffect of a Table 4.2 shows that initial obligees who are primary care physicians $ lOi Salary Increase by are most influenced by a $10,000 salary increase and those who are ob/ Gyn’s or in support medicine specialties are influenced the least. Medical Specialties Table 4.2: Estlmated Effect of a $10,000 Salary Increase on Various Specialists’ Internal Medical Probability of Leaving the Military (Initial Ob/Gyn Surgery Primary medicare support Obligdes) Average (mean) Military $78,568 $77,754 $68,768 $68,568 $76,061 pay Current probability of 90.0 86.5 72.5 82.9 90.7 leaving (percents) -___-__ Probability after salary 78.1 71.6 48.1 62.9 79.0 increase cDercents\ Decrease in probability 11.9 14.9 24.4 20.0 11.7 (percentage points) Page 28 GAO/HRD-90-1 Views on Military Medicine Chapter 4 Factors That Most Influence Physicians’ Plans About Military Service Table 4.3 shows that among other obligee specialists, internal medicine and medical support physicians appear to be the most influenced by a $10,000 salary increase. The predicted effects of this increase do not, however, differ very much between specialties. Table 4.3: Estimated Effect of a $10,000 Salary’lncreaae on Various Specialists’ Internal Medical Probe ility of Leaving the Military (Other Ob/Gyn Surgery Primary medicine support Oblige ! s) Average (mean) military - $94,861 $95,764 $80,842 $82,615 $87,783 pay ._.~_____ Current probability of 52.9 45.8 31.8 50.4 60.8 leaving (percents) .-____- Probability after salary 36.0 29.8 17.1 31.6 42.4 increase (percent) Decrease in probability 16.9 16.0 14.7 18.8 18.4 (percentage points) Responses to our survey also suggested that physicians’ intentions to Im@actof leave the military are affected by the number of hours they spend on Nonphysician Tasks nonphysician tasks. Physicians were dissatisfied with the number of on Stay/Leave administrative and health support personnel and the number of hours they spent performing nonphysician tasks. Eighty-seven percent of the Dedsion physicians who were dissatisfied with the number of administrative and health support personnel available indicated a 70-percent or higher probability of leaving the military. The regression analysis results confirm that the number of hours spent on nonphysician tasks has a statistically significant impact on both ini- tial and other obligees’ stated probability of leaving (see app. III). Spe- cifically, as the number of hours per week spent on nursing, clerical, and administrative support tasks increases, the stated probability of leaving also increases. Table 4.4 shows the estimated impact that nonphysician hours have on physicians’ intentions to leave the service. Specifically, the table shows how much the probability of leaving would decrease if nonphysician hours were reduced to zero. Page 29 GAO/HRD-90-l Views on Miitary Medicine Y Chapter 4 Factors That Most Influence Physicians’ Plans About Military Service Table 4.4: Estimated Impact of Eliminating Nonphysician Tar&k8on Initial Average number of hours per Decrease in probability of and Other Obligeer week spent on nonphysician Current probability of leaving (percentage tasks leaving (percent) points) 11.3 (Initial) 88.5 5.2 a.4 (Other) 52.7 9.5 For initial obligees, 863 physicians (of 1,000) reported working between 1 and 20 hours per week on nonphysician tasks. Of these, 764 would likely leave the military while 99 would likely stay. By eliminating non- physician tasks for physicians, an estimated additional 44 physicians would likely stay in the military-an increase over the original 99 of 44 percent. For other obligees, 869 physicians out of 1,000 reported spending between 1 and 20 hours per week on nonphysician tasks. Of this number, 466 are likely to leave the military while the remaining 413 are likely to stay. If nonphysician tasks were eliminated, an estimated 81 additional physicians would likely stay in the military-an increase over the original 413 of about 20 percent. By medical specialty group, the average reduction in the probabilities of leaving caused by eliminating nonphysician tasks ranges from 4.1 to 9.5 percentage points for initial obligees and from 7.6 to 9.8 percentage points for other obligees (see table 4.5). Table; 4.5: Estimated Impact of Elfminating Nonphysician Tasks on the Initial obligees Other obligees Probability of Leaving the Military, by Decrease in Decrease in MediCal Specialty Group Current probability of Current probability of Medical probabi;$;f probability of leaving specialty (perci~~ leavin CPerc;;;;f; wow (percent 7 (percent 7 ____- Ob/Gyn 90.0 4.0 52.9 -- 9.8 Surgery 86.5 5.9 45.8 9.5 Primary 72.5 9.5 31.8 7.7 Internal medicine 82.9 7.2 50.4 9.4 Medical support 90.7 4.1 60.8 9.0 Page 30 GAO/HRD-99-l Views on Miitary Medicine Chapter 4 Factora That Most Influence Physicians’ Plans About Military Service In our questionnaire we asked physicians whether it was difficult to Ability to Maintain maintain proficiency in their specialty given the number, diversity, and Prokiciency in Area of complexity of the cases encountered. Physicians who believed the case Spebialization mix encountered in the military made it difficult or very difficult to maintain proficiency in their area of specialization stated a higher probability of leaving the military. While only 40 percent of all physi- cians reported difficulty maintaining proficiency, the regression analy- sis shows a statistically significant relationship between probability of leaving and being able to maintain proficiency in a specialty. Specifically, the regression shows that, holding constant all other fac- tors influencing physicians’ intentions to leave, there is a significantly higher probability of leaving for those physicians who said that they had difficulty maintaining their specialty skills as compared with those who did not have any such difficulty. If it were possible to entirely elim- inate difficulty maintaining proficiency, initial obligees would be mini- mally affected in their decision to leave (a decrease of 1.2 percentage points in probability of leaving). The effect would be stronger on other obligees, (a 4,2-percentage-point decrease in probability of leaving) but this group reports a less than even chance of leaving the military. We estimated the effect of combat readiness training and field exercises Combat Readiness on the physicians’ stated probability of leaving. Results of this analysis Training indicated that for physicians serving under their initial obligation there was a statistically significant relationship between the number of hours spent on readiness training and the probability of leaving. Specifically, as the number of hours spent on readiness training and field exercises increases, the probability of leaving also increases. Reducing the number of hours spent in readiness training to 0, however, only changes the average probability of leaving by 1 percentage point. In any case, elimi- nation of combat readiness training for military physicians is not consis- tent with the primary mission of military medicine, which requires physicians to provide care for the sick and wounded in combat situations. We asked physicians how many permanent changes of station they con- Permanent Changesof sidered to be undesirable and we estimated the effect of this factor on Station their stated probability of leaving. The regression analysis results indi- Y cate that as the number of unwanted changes of station increases there is a statistically significant increase in the stated probability of leaving. Page 31 GAO/HRD-90-l Views on Military Medicine Chnpter 4 Factors That Must Influence Physicians’ Plans About Military Service This relationship is true for other obligees only. For example, eliminat- ing unwanted changes of station would reduce the average probability of leaving by about 3 percentage points. However, it would be difficult, if not impossible, for DOD to completely eliminate unwanted changes of station. 1 The probability of leaving stated by physicians is affected by several Sbary factors as shown in table 4.6. Tab14 4.6: impact of Potential Personnel Polic es on Retention of Military Figures are percentages Phys 1cians Initial obligees Other obligees --- __-- Current stated probability -___.__ of leaving 88.5 52.7 Percentage-point decrease in the stated probability of leaving__-___ when: Salary -_____- is increased -.- by $10,000 ___-- -_____ 14.3 18.8 -- Nonphysician II_- tasks are eliminated 5.2 9.5 Obstacles to maintaining proficiency are -~--- removed 1.2 4.2 - . --. Combat readiness training is eliminated 0.9 -~__ Unwanted permanent changes of station are eliminated . 3.2 Our analysis of these factors indicates that this probability can be reduced most substantially by increasing salaries and/or by decreasing hours spent on nonphysician tasks. Other factors, such as ability to maintain proficiency, unwanted changes of station, and readiness train- ing, show statistically significant effects but are impractical or difficult for DOD to address given its wide-ranging peacetime and readiness missions. Page 32 GAO/HRDSO-1 Views on Military Medicine Page 33 GAO/HRD-90-l Views on Military Medicine ,&&ndix 1 Physicians Planning to Leave the IMilitary Upon C@mpletionof Their Obligation This analysis contains statistics showing by service, by medical spe- cialty group, and by type of obligation the percentage of physicians stat- ing certain probabilities of leaving the service upon completion of their obligation. Two stated probabilities of leaving are shown, 70 percent or greater and 60 percent or greater. Table 1.1: Physicians Planning to Leave the Military Upon Completion of Their Obligation (By Service) Figur s are percentages Probability of leaving Air Force Navy Army All Physicians 70 getrcent or more 52 50 41 47 50 p¢ or more 63 65 59 62 Table 1.2: Physicians Planning to Leave the Military Upon Completion of Their Obligation (By Medical Specialty Group) Figures are percentages Surgical Internal support Ail Probabilitv of leavina Ob/Gvn specialties medicine Primarv care medicine phvsicians 70 percent , or more ..~_. - 60 51 46 42 71 47 50 percent or more 75 68 65 57 79 62 Table 1.3: Physicians Planning to Leave the Service Upon Completion of Their Obligation (By Type of Obligation) Figures are percentages Probability ., of -.leaving .._____ initial obligation Other obligation Ail physicians 70 percent or more 62 28 47 50 percent or more 77 42 62 Page 34 GAO/H&D90-l Views on Military Medicine Appendix II Physician Views on the Adequacy of Health and A’dministrative Support Personnel This appendix shows the variations among the five medical specialty groups regarding the adequacy of health and administrative support staff. We asked the physicians to specify whether they had more than, fewer than, or as many as needed of the type of staff required to sup- port their work. Physicians viewed shortages of support personnel dif- ferently because certain specialists require different types of health support personnel. However, all physicians indicated their are fewer administrative support personnel than they need. Table $1: Internal Medicine Physicians’ Vtews n Adequacy of Support Staff Figures are percentages (Percem)” Do not need Mor;;rt; Fewer than As yen;:; this tysa7: Type of staff needed Health support personnel Surgical nurses 0 23 1 75 Nurse anesthetists 0 17 3 80 Nurse practitioners 1 40 6 53 Nurse midwives 1 9 5 85 General dutv nurses 1 72 5 22 X-ray technicians 0 42 30 47 Corpsmen/orderlies 2 69 16 18 Physician assistants 2 36 7 55 Lab technicians 1 70 11 18 Administrative support personnel Administrative Officers 14 40 29 17 Medical Records Clerks 1 66 15 18 Secretaries, Clerks, Receptionists 1 91 8 0 Note: Percents may not add to 100 due to rounding Page 36 GAO/HRD90-1 Views on Military Medicine Appendix II Physician Views on the Adequacy of Health and Admtnistrative Support Personnel TabId II.2 Obrtetrlcian/Qynecologlsts’ Vie* on Adequacy of Support Staff Figures are percentages Do not need MzJ;r; Fewnert;i; As ye;ydtz this tyra;: Type of staff Health support personnel Surgical Nurses 1 74 23 2 Nurse Anesthetists 1 57 39 3 Nurse Practitioners 0 78 19 3 Nurse Midwives 2 60 16 22 General Duty Nurses 1 92 6 2 X-Ray Technicians 0 55 35 10 Corpsmen/Orderlies 0 84 14 2 Physician Assistants 2 36 20 42 Lab Technicians 0 77 15 9 Administrative support personnel Administrative Officers 26 26 44 4 Medical Records Clerks 0 77 21 1 Secretaries, Clerks, Receptionists 0 94 5 0 Note: Percents may not add to 100 due to rounding. Table 11.3:Primary Care Specialists’ Vle\irs on Adequacy of Support Staff Figures are percentages Do not need More than Fewnert;zi AS many as this tyza;; Type of staff needed needed Health support personnel Surgical Nurses 0 23 7 70 Nurse Anesthetists 0 17 11 72 Nurse Practitioners 1 44 17 38 Nurse Midwives 1 15 3 81 General Duty Nurses 1 71 11 17 X-Ray Technicians 1 42 30 28 Corpsmen/Orderlies 1 78 6 16 Physician Assistants I 49 15 35 Lab Technicians 1 56 25 18 Administrative support .. personnel . Administrative Officers 18 30 40 13 Medical Records Clerks 1 65 25 10 Secretaries, Clerks, Receotionists 2 84 12 3 Note: Percents may not add to 100 due to rounding Page 36 GAO/HRDBO-1 Views on Military Medicine Appendix II Phyeician Views on the Adequacy of Health and Adminbtrative Support Personnel Table Il.@:Medlcal Support Specialists9 Vlews or Adequacy of Support Staff Figures are percentages Do not need / Mo;;;ri; Fewn;;;z; As ye;%;; this tyza$ Type of staff Health support personnel , Surgical Nurses 0 46 9 45 I Nurse Anesthetists 3 34 17 47 Nurse Practitioners 1 13 6 80 Nurse Midwives 1 7 4 88 General Duty Nurses 1 52 4 44 X-Ray Technicians 1 58 16 24 Corpsmen/Orderlies 1 56 10 33 Physician Assistants 1 16 5 78 Lab Technicians 1 32 9 59 Administrative support personnel Administrative Officers 12 36 20 32 Medical Records Clerks 2 53 12 33 Secretaries, Clerks, Receptionists 2 87 9 3 Note: Percents may not add to 100 due to rounding Table 11.5:General and Other Surgical Specia(ist8’ Views on Adequacy of Figures are percentages Suppo* Staff Do not need MoWWe”; Fewnzrt.?rt~; As ye;%:; this tyza;f Type of staff Health support persofkel Surgical Nurses __- 0 74 11 15 -- Nurse Anesthetists 1 56 26 17 Nurse Practitioners 2 32 20 46 Nurse Midwives 2 15 10 73 General Duty Nurses 1 87 7 5 X-Ray Technicians -.- 1 62 24 13 Corpsmen/Orderlies 1 80 17 2 Physician Assistants 2 52 17 29 Lab Technicians -__-____ 1 61 26 13 Administrative support . personnel . Administrative Officers 31 24 38 7 Medical Records Clerks 2 68 25 6 Secretaries, Clerks, Receptionists 0 88 11 1 Note: Percents may not add to 100 due to rounding. Page 37 GAO/HRD-99-l Views on Military Medicine Appendix III Skvey and Data Analysis Methodology This appendix describes the sampling design and data collection tech- nique we used, as well as survey response rates. Because the data we report about military physicians are estimates that are based on a sam- ple of all physicians, this appendix also presents the sampling errors associated with these estimates. Further, this appendix describes the regression model we used to gain further insight into the factors that influence the probability of physicians leaving the military. We surveyed a random stratified sample of about 500 physicians from Shpling Design each branch of the service who were on active duty as of July 1988. We excluded all those whom the military considered interns or in graduate medical education programs because they were several years away from making a decision about their plans to leave or remain in the military. About 3,600 physicians were in these programs, leaving a remaining universe of approximately 9,600 physicians. We divided the remaining physicians into five groups, or strata, based on their military-designated medical specialty as shown below. Five Medical Strata and Examples of Medical Specialties Used in GAO Survey Ob$tetrics/Gynecology Sur@ons l Orthopedic Surgery . General Surgery l Neurosurgery Internal Medicine . Rheumatology l Gastroenterology . Endocrinogoly Primary care l General Medicine Y . Family Practice . Psychiatry Page 38 GAO/HRD-90-1 Views on Military Medicine AppendixIlI Survey and Dnta Analysis Methodology SuppcjrtMedicine . Anesthesiology . Radiology . Nuclear Medicine Officials from the services’ Offices of the Surgeons General and the Office of the Assistant Secretary of Defense for Health Affairs and GAO'S Chief Medical Advisor assisted us in grouping specialties. By stratifying our sample in this way, specialties with relatively few physicians were better represented in our overall sample than they would have been if we had drawn a simple random sample from each service. Table III.1 shows the total number of physicians in the universe in each stratum by branch of service, and the number of physicians we ran- domly sampled and surveyed from each stratum. Table IIf.1: Random Sample Selection- By Seryice and Stratum Number of physicians in Number of physicians universe sampled (As of July 1988) (As of July 1988) Stratum Army Navy Air Force Army Navy Air Force Ob/Gyn - 233 105 155 85 70 75 Surgeons 833 435 580 110 120 ___-- 120 Internal medicine 2,295 1,818 1,881 115 140 140 Primary care 558 153 215 105 80 85 Support medicine 298 210 210 85 90 85 Total 4,217 2,721 3,041 500 500 505 For each service, this sampling methodology allows us to project data from our survey to all active-duty physicians in each stratum as well as to all active-duty physicians. In general, estimates are for active-duty physicians who had completed their internship and were not graduate medical students as of July 1988. We developed a standardized mail questionnaire to collect information Questionnaire from our sample of physicians. We pretested the questionnaire with two Development physicians from each branch of the service. Based on the results of these pretests and comments from officials in Health Affairs and each of the services and the GAO Chief Medical Advisor, we revised the question- naire before mailing it out in November 1988. A follow-up mailing was ” conducted in December 1988. The questionnaire elicited information on several issues, including Page 39 GAO/HRD-90-l Views on Military Medicine Appendix Ill Survey and Data Analysis Methodology satisfaction/dissatisfaction with the number of hours worked; satisfaction/dissatisfaction with types and variety of patients treated; time spent on clinical, research, management, administrative, or other activities; . frequency with which physicians practice their primary specialty; . sufficiency/insufficiency of health, administrative, and clerical personnel; amount of time spent on quality assurance tasks; opportunities to attend professional conferences and seminars; satisfaction/dissatisfaction with compensation; and reasons for entering the military and probability of staying or leaving. I Of the 1,506 questionnaires we sent, 1,272 were returned, for an overall Su$vey Responses response rate of about 85 percent. Of the 1,272 respondents, 62 either indicated that they were no longer active- duty physicians or did not answer the question. Our overall study objective was to describe the experience, opinions, and attitudes of active-duty military physicians. Therefore, the estimates in this report are based on responses from the 1,220 known active-duty respondents. These respondents represent an estimated 9,616 active-duty physicians out of the 9,979 physicians in our original universe. Because we are estimating the characteristics of military physicians based on a sample of those physicians, there is an error or imprecision associated with each of these estimates. This imprecision is usually expressed as a sampling error at a given confidence level. Sampling errors for estimates from this survey were calculated at the 95-percent confidence level. For example, based on responses to our questionnaire we estimate that 90 percent of all military physicians are currently under some type of obligation to remain in the military. At a 95-percent confidence level, the sampling error for this estimate is plus or minus 2 percentage points. This means that the chances are about 19 out of 20 that the actual percentage of physicians currently under some type of obligation is between 88 and 92 percent. While some of the estimates in this report pertain to all military physi- cians, others are presented according to branch of service and/or spe- cialty group. At a confidence level of 95 percent, the sampling error for estimates for physicians across all services are generally less than plus or minus 4 percentage points. When estimates are for the physicians in the single service, sampling errors are generally plus or minus 7 percent- age points or less. Page 40 GAO/HRD-90-l Views on Military Medicine Appendix III Survey and Data Analysis Methodology Estimates for specialty groups generally have sampling errors of plus or minus 6 percentage points or less. Sampling errors for estimates involv- ing a specific specialty within a service do not exceed plus or minus 11 percentage points. Totals in tables throughout the report do not always add to the esti- mated universe of physicians (9,616) because not all respondents answered all the questions. Percent totals in tables do not always add to 100 due to rounding. To account for differences in physicians’ stated probability of leaving, Mulbiple Regression we developed a model that contains many factors plausibly related to Model and Analysis the probability of leaving. We estimated the model using multiple regres- sion, a standard statistical technique that quantifies the relationship between a dependent variable and a set of independent variables. Our dependent variable, described in detail on page 45, is closely related to the probability of leaving that physicians stated in their question- naire responses. The independent variables include such objective fac- tors as military and civilian compensation, medical specialty, the number of hours spent on nonphysician tasks, and hours per month in combat training, as well as proxies for tastes or preferences-gender, career plans at time of enlistment, and so on. The regression procedure related the variation between physicians in their stated probability of leaving to the variation in each of the independent variables, and esti- mated the effect on the dependent variable of a given change in each independent variable (holding the other independent variables constant). Based on the regression results, we calculated the change in the probability of leaving that would be associated with a given change in one independent variable, if the others remained unchanged. For exam- ple, we calculated the change in the probability of leaving associated with a $10,000 increase in military pay, assuming that the nonpay and personal characteristics variables remained unchanged. We also calcu- lated the change in the probability of leaving that would occur if the time physicians spend on nonphysician tasks were eliminated, assuming that the other independent variables (including pay) were held constant. Page 41 GAO/HRD-90-l Views on Military Medicine Appendix III Survey and Data Analysis Methodology The Model of Physicians’ After conducting bivariate analyses of our survey responses, we Prdbability of Leaving hypothesized that independent variables related to the physicians’ stated probability of leaving would fall into four categories: 1. Personal and demographic characteristics of the individual physician. 2. Military and civilian compensation. , 3. Noncompensation characteristics of the practice of military medicine. 4. Medical specialty. All the variables in the regression model and their definitions are listed in the following pages. Our four categories of independent variables resemble categories drawn from the standard economic theory of labor supply and occupational choice, these categories are: . compensation of military physicians and their compensation alternative in the civilian sector, l nonpay characteristics of both military and ,civilian medicine, and l tastes or preferences of the individual physician. Because the categories from economic theory are rigorously derived as well as comprehensive, they provide a benchmark for assessing the ade- quacy of our model.’ In addition, the theoretical categories provide a precise, economic interpretation of several of our independent variables. For example, the medical specialty variables in our model can be inter- preted as reflecting nonpay characteristics of military medicine that are included explicitly as independent variables-specifically, the working conditions peculiar to practicing a given specialty in the military.2 The personal characteristics variables provide another example: they can be ‘Two deviations of our model from the theoretical benchmark are noteworthy: first, economic theory dictates including nonpay characteristics of both the current occupation (military physician) and the alternative occupation (civilian physician), but we have data on nonpay variables only for the current (military) occupation. Second, economic theory calls for including a civilian pay variable that meas- ures-for each military physician-the compensation that would be earned by a civilian physician with the same specialty, training, work history, and ability as the military physician in question. Lacking such refined data, we used data on the average pay of civilian physicians in the same spe- cialty and at the same rank as the military physicians. (See p. 46 for more detail.) “The medical specialty variables may also reflect differences by specialty in civilian physician com- pensation, because our measure of that pay concept (CIVPAY) is not perfect. Page 42 GAO/HRD-90-l Views on Military Medicine Appendix III Survey and Data Analysis Methodology interpreted as proxies for differences in tastes that influence a physi- cian’s evaluation of the objective features, both pay and nonpay, that typify military medicine and civilian medicine. Definitions of Variables MALE = 1 if male Used!in RegressionModel MAR = 1 if married living together DPND = number of dependents CAR = 1 if planned military career when entered military MILPAY = In (military pay), where military pay is estimated by DOD spe- cialty and active duty grade CIVPAY = In (civilian pay), where civilian pay is estimated by DOD spe- cialty and academic rank NONPHYS = In (hours per week spent on nursing, clerical, and adminis- trative tasks) MIX = 1 if difficult to maintain proficiency with case mix PRSP = 1 if able to practice in specialty CT = hours per month spent in readiness or combat training DEP = average number of months deployed per year PCS = number of unwanted permanent changes of station. INT = 1 if specialty is an internal medicine specialty PRIM = 1 if specialty is a primary care specialty SURG = 1 if specialty is a surgery specialty OBGYN = 1 if specialty is obstetrics/gynecology SUPPORT MEDICINE used as reference group Persbnal and Demographic This category contains several demographic characteristics-gender (MALE), marital status (MAR), and number of dependents (DPND), as well Variables as a more direct indicator of personal preferences-whether the physi- cian planned on a military career when he or she entered the military (CAR). Military and Civilian We chose, as our measure of military pay, the average salary for mili- Compensation Variables tary physicians by specialty and grade level (MILPAY).~ The source for these average salary data is DOD. As our measure of the pay that alter- native positions in civilian medical practice might offer military physi- cians, we selected the average salary of teaching physicians by academic rank-assistant professor, associate professor, and full professor yi “MILPAYis expressedasthe natural logarithm of the dollar salary, as is our measure of civilian salaries (CIVPAY). Page 43 GAO/HRD90-1 Views on Military Medicine Appendix JlI Survey and Data Analysis Methodology - (CIVPAY). DODtreated these ranks (Health Professions Special Pays Study) December 1, 1988) as comparable to military physicians at pay grades 04,06, and 06 (Army and Air Force: Major, Lt. Colonel, and Colonel; Navy: Lt. Commander, Commander, and Captain). We excluded pay grade 03 physicians from our regression analysis, because we lacked DOD data on physicians with active duty grade 03 (Army and Air Force: Captain; Navy: Lieutenant). Academic salaries are not a perfect substitute for earnings in the private sector (which has nonteaching physicians in clinical and administrative roles). In addition, teaching medicine may not be the typical civilian alternative for the military physician. Nonetheless, we used academic salaries as a proxy for civilian earnings for two reasons. First, academic salaries were the best civilian salary data readily available that per- tained to civilian positions comparable to military physicians’ duties and experience. Second, review of research on military physicians’ retention and pay suggests that academic salary information provides a measure of civilian physician compensation that is roughly as good as informa- tion on nonacademic salaries. Specifically, a Congressional Budget Office study, which used information collected by the American Medical Asso- ciation on both academic and nonacademic salaries, found effects of civilian pay on military physicians’ attrition that are similar to a DOD study that used data only on academic salaries. Although information is lost by using data only on academic salaries, evidently the academic data vary sufficiently with civilian physician pay in general to permit meaningful analysis of the relationship of pay to military physicians’ retention.4 Nonpay Variables For the category of nonpay characteristics of the job of military physi- cian, we chose variables that either are associated exclusively with the practice of military medicine or apply to anyone in the military. These variables include the adequacy of support personnel, measured by the logarithm of the number of hours spent on nonphysician tasks (NONPHIS); adequacy of case mix (MIX); whether physicians are able to 4Measurement errors, such as those affecting our civilian pay indicator, may cause the coefficient estimate on CIVPAY to be biased toward zero. Also, this so-called “errors in variables” problem also causes coefficient estimates to be less statistically precise than otherwise. This problem may explain in part why the coefficient on CIVPAY is insignificant in the other obligees equation and why the coefficient for civilian pay is substantially smaller than that for military pay in all equations. Accu- racy in civilian pay may be more important for other obligees than initial obligees, because the civil- ian opportunities for more experienced physicians (other obligees) may be more heterogeneous than for less experienced physicians (initial obligees). Page 44 GAO/HRD-90-l Views on Military Medicine Appendix III Survey and Data Analysis Methodology practice in their area of specialty (PRSP); hours per month spent in mili- tary readiness (i.e., combat) training (CT); average number of months deployed per year (DEP); and number of unwanted permanent changes of station (PCs). The I)ependent Variable We formulated the dependent variable as the natural logarithm of the and qrobability of Leaving odds ratio, (Prob/(l-Prob)), where “Prob” denotes the stated probability of leaving. The probability of leaving would be more intuitive as the dependent variable, but it is prone to a statistical problem: a regression model may predict a probability of leaving that exceeds 1 or is less than O-a result that contradicts the meaning of probability. To avoid this difficulty, it is conventional to define the dependent variable as the loga- rithm of the odds ratio. Initial and Other Obligees We performed the regression analysis separately for physicians serving Analyzed Separately under their initial obligation (initial obligees) and for those serving under some other form of obligation (other obligees). Analysis of the physicians’ questionnaire responses revealed that these two groups stated substantially different probabilities of leaving the military (on average). In addition, the two groups are likely to have, on average, different tastes or preferences for military life and military medicine, objective factors (pay, etc.) aside. Other obligees have, by accepting one or more obligations beyond their initial obligation, revealed a preference for mili- tary life and military medicine.6 By contrast, the group of initial obligees contains both those physicians who will eventually incur a second obli- gation and those physicians who will, when their initial obligation is up, leave the military. Consequently, it is reasonable to assume that the ini- tial obligees have a lesser taste for military medicine, on average, than other obligees have. In addition, initial obligees and other obligees may evaluate the particular objective features of military medicine (e.g., unwanted permanent changes of station) differently. Given these con- siderations, economic theory suggests that the two groups be analyzed separately.” “At the end of their initial obligation, other obligees faced the same pay and nonpay characteristics of military medicine as their colleagues who chose to leave the military. This strongly suggests that other obligees have a greater taste for military life and military medicine than those who leave the military. “In economic terms, if the preference functions of the initial obligees differ from those of other obli- gees, then the derived functions for probability of leaving will differ between the two groups as well. Page 45 GAO/HRD-90-l Views on Military Medicine Appendix lI.I Survey and Data Analysis Methodology Results of Regression We estimated the regression models for initial and other obligees by the Analysis method of ordinary least squares. Table III.2 presents the estimates of the regression coefficients for the two models of initial and other obli- gees’ intentions to leave the military. The table also contains informa- tion on the statistical precision (standard error and T-statistic) of each coefficient estimate. An estimate is considered statistically significant if the probability is low that the true value of the coefficient is 0. We chose as our criterion a significance level of .lO; that is, we required that the probability of the true coefficient being 0 is no greater than .lO. Tablei 111.2:Estimates of Regression Equations for Initial and Other Obligees Depehdent ._. Variable=LN(Prob/( .._._-/-.“._-. ..___. “.__..--- 1-Prob)) Coefficient Standard Error T-Statistic Mean Indeiendent variable Initial Other Initial Other Initial Other Initial Other MILPdY Military pay -7.71 -6.91 4.98 2.01 -1.55 -3.45 11.22 11.37 ClVPpiY _ _ ..-., Civilian pay .^.._-__--..- 1.33 0.87 0.98 1.16 1.35 0.75 11.51 11.68 NONPHYS Nonphysician hrs. 0.07 0.06 0.04 0.03 1.82 1.98 1.64 0.90 MIX Case mix 0.26 0.44 0.20 0.22 1.30 2.00 0.44 0.38 PRSP’Practice specialty -0.26 -0.17 0.27 0.35 -0.96 -0.50 0.83 0.89 .._ _.,.__. ...-__~.-- CT Rqadiness training 0.02 0.00 0.01 0.01 1.90 -0.25 4.10 3.94 DEP deployments __~ 0.01 -0.14 0.12 0.13 0.10 -1.12 0.29 0.36 PCS Permanent chanae of station 0.12 0.14 0.12 0.09 1.02 1.45 0.46 0.94 MALE..,__._ Male-.-.- ._.....-. 0.01 -0.42 0.31 0.45 0.03 -0.95 0.86 0.94 _MS Marital status -0.14 -0.34 0.29 0.34 -0.47 -1 .Ol 0.85 0.87 DPT -...--No.. ..._....___^_ ..._._-- dependents- --_-._-..--_ --- 0.01 0.05 0.07 0.07 0.18 0.76 3.34 3.94 CAR Planned militarv career -0.74 -0.10 0.25 0.29 -2.97 -0.35 0.19 0.17 INT ln$ernal medicine -0.97 -0.62 0.47 0.39 -2.08 -1.57 0.18 0.24 PRIM ‘Primary care -1.62 -1.52 0.51 0.43 -3.19 -3.51 0.10 0.24 .-- _~- ..- SURG Surgery ..-.-_ ._,._.. -----II- -0.41 -0.07 0.28 0.35 -1.47 -0.20 0.29 0.23 OB Ob/Gvn 0.29 0.35 0.36 0.44 0.79 0.78 0.20 0.13 Constant 73.47 69.07 48.51 12.02 1.51 5.75 Adjusted ...____.._ R-square .._ ._..----- 0.08 0.13 F-Statistic 2.61 4.78 Sianifrcance level of F-Statistic 0.001 0.000 No. of observations 294 405 Note: “Prob” denotes physician’s stated probability of leaving military service. The adjusted R-square measures the proportion of the variation in the dependent variable that is accounted for by the variation in the set of explanatory variables. The F-statistic permits testing for the statistical significance of the observed association between the dependent variable and the set of explanatory variables. The sign of the coefficient indicates the direction of the estimated effect. For instance, the negative sign on MILPAY indicates that an increase in Page 46 GAO/HRD-90-l Views on Military Medicine , Appemiix III Survey and Data Analysis Methodology military pay results in a decrease in the stated probability of leaving. The “T-statistic” permits the statistical significance of the effect to be tested, Roughly, if the T-statistic if greater than 1.65 (1.28 for one-tailed tests), the effect is considered “statistically significant” at the lo-per- cent level. As the T-statistics in table III.2 reveal, three variables-military com- pensation, number of hours spent on nonphysician tasks, and case mix-have a statistically significant effect on the dependent variable (and, therefore, on the probability of leaving) for both initial and other obligees. Military pay is statistically significant, and has the expected negative effect on the physicians’ stated probability of leaving.7 That is, as military pay increases (holding civilian pay and other variables con- stant), the stated probability of leaving decreases. The number of hours per week that a physician spends doing nursing, clerical, and adminis- trative tasks has a statistically significant, positive coefficient. As non- physician hours (NONPHE) increases, the stated probability of leaving increases. Finally, the coefficient on case mix is statistically significant and positive. Physicians who said that they had difficulty maintaining proficiency in their specialty also stated a higher probability of leaving.R The regression results also indicate that initial obligees’ intentions to leave are affected by several variables that do not affect other obligees’ intentions. These variables are: hours spent on readiness training (CT) (positive effect), civilian pay (positive effect), and “planned a military career at time of enlistment” (CAR) (negative effect). One variable that does not affect initial obligees’ intentions does, how- ever, influence the intentions of other obligees. That variable-the number of undesired changes in permanent station-has a negative effect. Issues of Interpretation The multiple regression methodology that we have employed has the virtue that it provides quantitative answers to questions that are of cen- tral importance to policy regarding retention of military physicians. In particular, the survey data and regression estimates presented in chap- ter 4 and in this appendix provide guidance on the effects of military 7The appropriate statistical test is one-tailed, because the alternative to the null hypothesis of no effect was that the true coefficient is negative. sin addition to these three variables, the set of medical specialty variables is statistically significant for both initial and other obligees. Page 47 GAO/HRD-90-l Views on Military Medicine Appendix III Survey and Data Analysis Methodology pay. This guidance is the best we can provide given the general state of knowledge and the resources devoted to the study. As in any similar study, however, care and caution are called for in the interpretation and application of our results. This section discusses the principal qualifica- tions and problems of interpretation that users of our study should be aware of. First, it should be emphasized once more that our dependent variable relates to the stated probabilities of leaving the service that military physicians expressed in response to our survey. The overall picture pro- vided by their responses is reasonably consistent with actual experi- ence, and many specific features of our results-such as the higher separation probabilities indicated by initial obligees-are quite plausi- ble. Nevertheless, we have not established a direct link between the probabilities stated in the survey results and the actual decision of mili- tary physicians to stay in the service or not. Second, our estimates of the effects of the various independent variables are relatively imprecise. In particular, although we have established, by conventional statistical standards, that military pay does affect physi- cian retention, the standard errors reported in table III.2 indicate that we have not succeeded in measuring the strength of these effects with precision. For example, it is roughly an even odds bet whether the esti- mated value of the MILPAY coefficient for other obligees (-6.91) is within +- 1.36 of the true value. Nonetheless, in our view, the figures presented in chapter 4 on the effects of a $10,000 salary increase are a reasonable basis for policy decisions. They do not, however, obviate the need for further study or for close monitoring of the results if an attempt should be made to address the retention problem through a salary increase. Third, it is possible that our estimates are systematically biased away from the true values. We have already noted that the explanatory vari- ables provided in our data set are less than ideal from the viewpoint of economic theory (page 44, footnote 1). Regarding civilian pay, we have explained the sort of bias that might result (page 46, footnote 4). Other possible sources of biased estimates might be adduced. We believe, how- ever, that any theoretically plausible analytic framework would yield similar results if applied to the same data or to better and more exten- sive data of the same type. In assessing the confidence of our conclusions, we paid particular atten- tion to the following plausible argument suggesting that our estimates of the effect of military pay might be too high: physicians whose tastes Page 48 GAO/HRD-90-l Views on Military Medicine Appendix III Survey and Data Analysis Methodology lead them to prefer military practice and life to the civilian alternative tend both to remain longer in the military and to report lower probabili- ties of leaving, compared with physicians who lack such a preference. Because military pay generally rises with years of service, military pay and “taste for military life” would mean an estimate of the pay effect that is too high; part of our measured effect of higher pay would be attributable to a stronger taste for military life. Before evaluating this argument, we note that it illustrates a general point about the type of analysis we have done. The Congress could raise pay, but it presumably cannot raise physicians’ taste for military life at the same time. It is precisely for this reason that we employed multiple regression analysis to attempt to assess (among other things) what the effect would be if military pay and nothing else were changed. In pursu- ing this objective we attempted to control for taste effects by (1) divid- ing our sample into two categories (initial obligees and other obligees) that are very likely to differ on average in their taste for military life, and (2) introducing a variable (CAR) into the model that indicated whether a physician, upon enlistment, intends to pursue a military career. Recognizing that these efforts to control for taste factors might have fallen short, we performed further analysis of the data on other obligees to determine whether years of service in the military might have some- thing to do with the relationship between pay and retention. We did find some suggestions in the data that the physicians with more than 12 years of service (roughly one-half the sample) have lower stated probabilities of leaving and are less responsive to pay than physicians with fewer years of service. However, these indications are not so strong as to lead us to reject at standard significance levels the hypothesis that there is no difference between the groups with high and low years of service. In addition, although physicians with more years of service col- lectively report lower separation probabilities on average, they are quite comparable to their colleagues in terms of the (large) amount of varia- bility around that average. Thus, it is far from the case that all long- tenured military physicians have simply decided that they are staying in the service. Also, physicians with fewer years of service appear to be, if anything, more responsive to military pay effects than we have esti- mated. Overall, it seems clear that our estimate of the pay effect is not merely an artifact associated with taste variables for which years of service provides a proxy. And while it remains plausible that long-ten- ured physicians are less responsive to pay than short-tenured physi- cians, our data do not provide decisive support for this view. Page 49 GAO/IiRIHO-1 Views on Military Medicine p&ndix IV &&r Contributors to This Report David P. Baine, Director, Federal Health Care H I an Resources T Diyision, Delivery Issues (202) 276-6207 Stephen P. Backhus, Assistant Director Wzjshington, ! DC. William A. Hightower, Assignment Manager M. Cristina Rey Gobin, Evaluator-in-Charge Wayne M. Dow, Statistician Clarita A. Mrena, Social Science Analyst Jonathan Ratner, Assistant Director, Economic Analysis James Cosgrove, Economist (101343) Page 50 GAO/HRD-90-l Views on Military Medicine
Defense Health Care: Military Physicians' Views on Military Medicine
Published by the Government Accountability Office on 1990-03-22.
Below is a raw (and likely hideous) rendition of the original report. (PDF)