Objective. To determine factors influencing career choices by pediatric residents and how they may change the future pediatric physician workforce.
Background. Societal norms and policy decisions can influence the demographics of residents entering pediatric practice and resident career choices. Although predictors of physician career choice have been identified retrospectively by examining the characteristics of physicians in practice, little work has focused on the job selection factors important to pediatric residents when they make their postresidency career decisions.
Design/Methods. For 3 consecutive years (1997–1999), the American Academy of Pediatrics surveyed a national sample of 500 third-year residents, totaling 1500 residents. Data on resident characteristics, job selection attitudes, and career decisions were obtained, and trends for the past 3 years were examined.
Results. The percentage of residents intending to practice primary care remained constant at around 69% from 1997 through 1999. Female residents, US medical graduates, and residents married to nonphysicians were more likely to report primary care practice as their future clinical goal. These relationships were mediated by higher rates of a primary care practice goal among those who felt geographic location and future colleagues were very important and by lower rates among those who felt acceptable income, teaching opportunities, and research opportunities were very important. Six percent of residents entering general pediatrics were heading to rural locations.
Conclusions. With a projected increase in the number of female pediatricians and a decline in international medical graduates, our study suggests that pediatrics may continue to shift toward an increased proportion of general pediatricians. Lifestyle issues are a major factor influencing job choice and must be addressed.
- pediatric manpower
- physician supply and distribution
- physician practice patterns
- women physicians
- international medical graduates
Achieving an appropriate pediatric physician workforce is a critical element to providing health care to the children of our country. Maldistribution of pediatricians by specialty, geography, and ethnicity can limit children’s access to quality care. Our society invests tremendous resources into physician education, and the long training period for physicians makes rapid changes difficult. To ensure that children will be able to access pediatric care in the future, it is imperative that factors of importance to residents when making career choices are identified and trends in the pediatric workforce are anticipated.
Driven in large part by changing societal norms and policies, several demographic trends are currently influencing the pediatric workforce. First, for several years women have constituted the majority of physicians entering pediatric residencies. As a result, the percentage of women in the pediatrician workforce is increasing and was 47.5% in 1999.1,2 Moreover, women are affected more by family factors when making career choices than men.3–6 Second, there is a recent decline in the percentage of international medical graduates (IMGs) entering pediatrics, resulting from increased interest in pediatrics among US medical graduates (USMGs), and increasing training and job restrictions placed on IMGs in the past decade.7 Third, the racial and ethnic diversity of medical students has been constrained because of the elimination of affirmative action in higher education institutions in several states8 and reduction of educational programs that actively recruited minority medical students.9 Simultaneously, the diversity of children in the United States is dramatically increasing.10
Female gender, underrepresented minority status, and IMG status have all been linked to physicians’ decision to practice primary care medicine. Whereas females and underrepresented minorities (URMs) are more likely to practice primary care medicine, IMGs are more likely to subspecialize.1–13 IMGs are also more likely to practice in urban inner-city1 and rural15 locations immediately after residency. Other resident characteristics, such as marital status and educational debt, have also been examined in relation to the choice to practice primary care. Specifically, married physicians are more likely to practice primary care, and educational debt has not shown any relationship with the choice to practice primary care.11,13
Although most previous studies examined the current distribution of pediatricians and the career choices of medical students, we seek to provide insight into the evolution of the pediatric workforce by examining the characteristics and choices of pediatric residents who are at the “leading edge” of the pediatric physician workforce. We will examine resident career intentions, job preferences, and postresidency job choices and make comparisons with various resident characteristics.
A national random sample of 500 third-year categorical or primary care pediatrics residents was surveyed by the American Academy of Pediatrics during their last months of training (April to August) in the years 1997, 1998, and 1999, for a total sample of 1500 residents. The residents were randomly selected from a database of 7911 residents enrolled in the Pediatrics Review and Education Program conducted by the American Academy of Pediatrics. Pediatrics Review and Education Program is an educational program offered free to all pediatric residents in Accreditation Council for Graduate Medical Education-accredited US pediatric residency programs. A total of 1205 (80%) residents returned the survey. The response rates were 78% in 1997, 84% in 1998, and 79% in 1999. The demographic characteristics of responders and nonresponders, including age (mean years: 31.6 vs 31.9; P = .27), gender (percentage of females: 63 vs 59; P = .19), and medical school location (percentage of IMGs: 28 vs 29; P = .72) did not differ significantly.
The survey asked residents about their education and training, their career goals, their postresidency positions, and factors influencing their job selection. Demographic information about the respondents’ race and ethnicity, gender, and medical school was also collected. Four mailings of the survey were mailed each year. Once residents responded they did not receive additional mailings. As an incentive to participate, the respondents were included in a raffle for a medical textbook of their choice. The survey varied slightly from year to year, and this study focuses primarily on questions that were common across all 3 years.
Comparisons between residents were made based on several resident characteristics including gender, medical school location (USMG or IMG), underrepresented minority race or ethnicity, marital status, parental status (have children or not), and higher educational debt (≥$50 000). Residents who were Hispanic, black, or Native American USMGs were classified as URM. The outcome variables of interest included residents’ postresidency job position, the location of residents’ general pediatrics position, importance of several job selection factors, and residents’ future clinical practice goal.
χ2 analyses were used to examine bivariate relationships between resident characteristics and outcome variables. Some values were missing for each of the predictor or outcome variables, producing slight variation in the number of cases used in each analysis. A P value <.05 was considered statistically significant for all analyses. In addition to describing the individual associations between resident characteristics and outcome variables, a multivariate hierarchical logistic regression model was developed to predict the characteristics of residents whose future clinical goal is primary care practice. Initially, all predictor variables were screened based on their bivariate relationship with the primary care goal outcome variable. Predictors with a P value >.20 were excluded from additional analysis. Remaining predictor variables were then entered into the multivariate model in 2 stages. In stage 1, all resident characteristic variables that survived screening were entered simultaneously and those that were significant (P value < .05) were retained. Next, in stage 2, the job selection factors that survived screening were added to the significant resident characteristics. The significant job selection factors were then retained. To demonstrate how the influences of resident characteristics on primary care decisions were, in turn, explainable by residents’ job selection attitudes, all significant predictors from stage 1 were kept in stage 2 regardless of whether they were nonsignificant at stage 2. All categorical coding variables were included in the model when any individual categorical coding variable was significant. Overall model performance was assessed using the c-index or area under the receiver operating characteristic curve.16
In Table 1, the number of responses is reported in the first column for each of the demographic characteristics. Of respondents, 63% were female, 27% were IMGs, 7% were URMs, 70% were married, 38% had children, and 48% had educational debt, including spouse debt, that exceeded $50 000.
Table 1 also shows the postresidency positions chosen by residents. Overall, 56% were heading into general pediatrics practice, 21% were entering a subspecialty fellowship, 10% were becoming chief residents, 9% had no position when surveyed, and 4% took some other position after residency. Different patterns of postresidency positions were found for several of the resident characteristics, including gender (χ2 = 14.4; P = .006), IMG status (χ2 = 42.8; P = .001), marital status (χ2 = 27.8; P = .001), having children (χ2 = 13.7; P = .008), and having higher debt (χ2 = 29.8; P = .001). Women were significantly more likely than men to take a general pediatrics practice position (58% vs 53%). IMGs were less likely than USMGs to take a general pediatrics practice position (44% vs 61%). Residents married to a nonphysician were more likely than unmarried physicians to take a general practice position (62% vs 47%), as were residents with children compared with those without (62% vs 52%) and residents with higher debt compared with those with lower debt (61% vs 52%).
Among residents entering general pediatric practice, most (52%) will be working in a suburban location, 26% will be working in an urban noninner-city location, 16% will be working in an urban inner-city location, and 6% will be working in a rural location (Table 2). This question was not asked of residents completing training in 1997 and thus the total number of respondents is lower for this table. Significant differences in location were found for IMG status (χ2 = 21.1; P = .001), and higher debt (χ2 = 10.2; P = .017). IMGs were more likely than USMGs to practice in rural or urban inner-city locations (40% vs 18%), and residents with lower debt were more likely to practice in rural or urban inner-city areas than residents with higher debt (27% vs 18%).
In addition to tracking residents’ current job selection, we also examined residents’ future clinical practice goal. This allows for a better approximation of the total number of residents eventually entering primary care pediatrics because it includes chief residents and those with fellowship training plans who may plan on an eventual primary care career.17 The first column of Table 3 shows that, overall, 69% of residents had a primary care future practice goal. Residents were considered to have a primary care practice goal only if that was their exclusive practice goal; thus, residents with an interest in combined primary care and subspecialty care practice were not considered to have a primary care practice goal. The relationship of residents’ future practice goal with each of the resident characteristics is very similar to the relationships that were seen for general pediatric practice positions in Table 1. Women residents (73% vs 61%; P = .001), married residents (73% and 69% vs 61%; P = .001), residents with children (72% vs 66%; P = .026), and residents with high debt (75% vs 63%; P = .001) were more likely to have primary care as their future practice goal. IMGs were less likely to have primary care as their future practice goal (58% vs 73%; P = .001). Unlike in Table 1, the proportion of URMs with primary care as their future practice goal was significantly greater than that of non-URMs (81% vs 68%; P = .011).
Residents were asked to rate various job selection factors related to careers in pediatrics, using a 4-point Likert scale (essential, very important, somewhat important, and unimportant). Table 3 reports the percentage of residents who rated a factor as either “essential” or “very important.” The factors are ordered in overall importance from left to right. Factors with the highest percentages of importance ratings were spouse/family considerations (90%), job security (87%), and geographic location (84%). Factors with low percentages were patient population (57%) and teaching opportunities (52%). Just over one fifth of residents considered research opportunities an important factor when making their career choice (21%).
Residents’ job selection ratings varied based on several resident characteristics. In comparison to men, more women reported that geographic location (86% vs 79%; P = .005), future colleagues (85% vs 78%; P = .004), and control over working hours (86% vs 75%; P = .001) were very important, and fewer women compared with men rated teaching opportunities (48% vs 59%; P = .001) and research opportunities (17% vs 28%; P = .001) as very important. In comparison to USMGs, IMGs had higher rates of reporting job security (95% vs 85%; P = .001), acceptable income (91% vs 79%; P = .001), and research opportunities (31% vs 18%; P = .001) as very important, and lower rates of reporting that geographic location (72% vs 88%; P = .001) and future colleagues were very important (74% vs 86%; P = .001). Compared with non-URM residents, URM residents had higher rates of reporting that their patient population was very important (81% vs 55%; P = .001). Compared with residents without children, residents who are parents had higher rates of reporting that family considerations (96% vs 87%; P = .001) and income (85% vs 80%; P = .020) were very important, and lower rates of reporting that future colleagues (80% vs 84%; P = .046) were very important. Compared with residents with lower debt, residents with higher debt had higher rates of reporting that geographic location (88% vs 79%; P = .001), future colleagues (85% vs 80%; P = .031), and patient population (60% vs 54%; P = .037) were very important and a lower rate of reporting that job security (85% vs 90%; P = .007) or research opportunities (17% vs 25%; P = .002) were very important.
The importance of the job selection factors also varied by marital status. Unmarried residents had higher rates of reporting that research opportunities were very important (27% vs 23% and 17%; P = .002) and lower rates of reporting that family considerations (75% vs 96% and 98%; P = .001), future colleagues (compared with residents married to a physician, 79% vs 89%; P = .004), or control over work hours (compared with residents married to a physician, 77% vs 86%; P = .009) were very important.
Although several resident characteristics were associated with residents’ future clinical practice goal, the bivariate analyses could not address the amount of similarity or shared variability between resident characteristics. Accordingly, these characteristics were analyzed together in a 2-stage multivariate logistic regression model to examine the independent effects of these variables. Resident characteristics were entered together in stage 1, and resident characteristics plus job importance factors were entered together in stage 2. Primary care practice as the future career goal was the outcome variable of interest. Three resident characteristics were found to be significant in the stage 1 multivariate model. Women were significantly more likely than men to report primary care as their career goal (odds ratio [OR]: 1.74; 95% confidence interval [CI]:1.32–2.29). IMGs were significantly less likely than USMGs to be interested in primary care (OR: 0.58; 95% CI: 0.43–0.78), and residents married to a nonphysician were more likely than unmarried physicians to be interested in primary care (OR: 1.93; 95% CI: 1.41–2.63).
Considerably more variability in the future practice goal outcome was explained when the job selection factors were entered into stage 2 of the model. The model c-index, which ranges from 0.50 to 1.0, increased from 0.63 to 0.79. The ORs moved closer to 1.0 for all of the resident characteristics in stage 2 (gender, OR: 1.38; IMG, OR: 0.87; married to nonphysician, OR: 1.61) suggesting that residents’ job selection attitudes partially explain the clinical goal differences between resident groups. In fact, gender and IMG status were no longer statistically significant when job selection factors were included in the model. By far, the strongest predictor of residents having a primary care practice goal was the importance of research opportunities. Residents who thought research opportunities were very important were much less likely to have a primary care practice goal (OR: 0.11; 95% CI: 0.07–0.16). An OR of 0.11 is comparable with a positive OR of 9.49. A primary care future practice goal was also negatively related to reporting that acceptable income was very important (OR: 0.64; 95% CI: 0.41–0.99) and that teaching opportunities were very important (OR: 0.66; 95% CI: 0.46–0.94). Finally, residents who thought geographic location (OR: 2.2; 95% CI: 1.44–3.35) and future colleagues (OR: 1.8; 95% CI: 1.19–2.72) were very important were more likely to have a primary care practice goal.
This study provides new insight into the future pediatric physician workforce by surveying pediatric residents about their career intentions and preferences at a time when such decisions would be uppermost in their mind. American Board of Pediatrics estimates show that there was an average of 2490 categorical pediatrics residents completing training each of the survey years. With 56% of residents entering general pediatric practice after residency and 69% intending a career practicing primary care (includes chief residents), we project that each year 1394 residents enter general pediatrics practice and 1718 residents intend to enter primary care practice eventually. The direct-entry rate is roughly 10 percentage points higher than a 46% estimate by residency program directors in 1990.18 Although increasing the proportion of trainees entering primary care has been a national policy goal, there is great concern in the pediatric academic community about decreasing numbers of residents pursuing subspecialty training and research.19
Extrapolating from the geographic location data we collected, only 6% of residents entering general pediatrics practice or 84 residency graduates each year will practice in a rural area. In 1998 there were 1242 counties in the United States that had a family practice or general practice physician but did not have a general pediatrician.20 Thus, it is unlikely that geographic maldistribution of pediatricians will change in the near future. Our findings suggest that lifestyle issues, which rank highest among the job selection factors, may need to be addressed to attract pediatricians to underserved areas. More research is needed to resolve this critical problem.
Our findings further indicate that the demographic characteristics of residents may greatly impact the balance of general pediatricians and pediatric subspecialists in the workforce. Perhaps, most important are the differences between men and women pediatricians. Our data show that women are more likely than men are to choose careers in primary care pediatrics. These differences, however, are actually based on differences in the importance of particular job selection factors including geographic location, control over working hours, future colleagues, and teaching and research opportunities. Our study suggests that policies designed to increase female pediatricians’ interest in teaching and research or designed to better accommodate pediatricians’ control over work hours and family needs, such as providing part-time opportunities, may increase the number of women entering subspecialty fellowships. It is notable that family considerations ranked as the most important job factor for both women and men, thus these concerns are not specific only to women.
Residents married to nonphysicians were also more likely to enter primary care than nonmarried residents. Because 56% of the residents married to nonphysicians and 42% of the residents married to physicians reported having children, the family demands facing married residents may be much greater than those facing nonmarried residents. Thus, these residents may avoid fellowship training to have greater control over their working hours.
IMG status and the employment policies pertaining to them also are likely to affect the balance of general pediatricians and pediatric subspecialists. Compared with USMGs, IMG residents seem to be more subspecialty-oriented with a greater interest in research. Additional differences between IMGs and USMGs that were predictive of a primary care practice goal included greater interest among IMGs in acceptable income and lower interest in geographic location and future colleagues. IMGs also rated job security as very important more often than USMGs, but this factor was not significant in the multivariate model. Thus, IMGs seem to consider potential economic vulnerability in their job selection more than USMGs do. In addition, our results show that IMG residents are more likely to not have a position at the conclusion of residency and to locate in potentially underserved, urban inner-city and rural areas immediately after residency. However, previous studies indicate that IMGs may not remain in these areas after they establish their practices.21
URM pediatric residents were more likely to have primary care as a practice goal than non-URM residents, and they were also more concerned about the characteristics of their patient population than non-URM residents. Overall, however, there was a very low percentage of URM residents in our study, which may contribute to the lack of significance of this factor in the multivariate model. On a larger scale, the low percentage of URM pediatricians being trained is a problem for the pediatrician workforce, because URM physicians are more likely to care for minority and disadvantaged patients.22 Based on the American Board of Pediatrics average of 2490 pediatrics residents completing training each year, we estimate that only 5% of residents or 125 residents are URMs entering practice and only 1% of residents or 25 residents are URMs entering fellowship training.
The influence of educational debt on the workforce is less clear. Consistent with previous research on pediatric subspecialization,11 educational debt was not a significant predictor of a primary care practice goal in the multivariate model, although in the bivariate analysis residents with higher debt were more likely to have primary care as a practice goal. Residents with higher debt were also less likely to choose to practice in urban inner-city and rural locations. Debt may be a barrier for some residents to choosing to practice in underserved areas and or academic careers, in favor of more lucrative practice opportunities in the suburbs.
One limitation to our study is that we obtained information from residents around the time of residency completion, and therefore the information is based on job expectations, and not actual experience. Initial work experience may change their career expectations and future plans. A second limitation is that, despite accumulating 3 years of data, the numbers of URM residents remain small, and there may be important differences that the survey did not have the power to detect. Similarly, the very small percentage of residents entering rural practice prevented us from examining this very important issue further.
Our examination of the “leading edge” of the pediatric workforce indicates that current trends may impact the health care needs of children in the future. Women will soon comprise a majority of pediatricians, yet there are societal and professional barriers that may inhibit women from choosing pediatric subspecialty careers and entering practice in underserved areas, especially rural areas. These barriers not only affect women, but men are also giving family considerations more importance in their job choices. Solutions to resolving specialty and geographic maldistribution must address lifestyle issues such as flexible and part-time work arrangements; child and elder care benefits; and career opportunities for spouses in underserved regions. Expected declines in IMG numbers with stricter immigration policies in place require greater efforts to make pediatric subspecialty careers more attractive to USMG residents to maintain a balance of general pediatricians and subspecialists. Finally, the widening gap between the diversity of the pediatric workforce and the diversity of children in our country must be addressed through increasing educational opportunities for minorities. Thus, the “leading edge” of the pediatric physician workforce challenges the pediatric community and policy makers to address the needs of future pediatricians so they can meet the needs of tomorrow’s children.
INCREASED SPENDING ON DRUGS IS LINKED TO MORE ADVERTISING
“The 50 most-advertised prescription medicines contributed significantly last year to the increase in the nation’s spending on drugs...The report was prepared by the National Institute for Health Care Management, a nonprofit research foundation that was founded by the Blue Cross Blue Shield health insurance plans.
Increases in the sales of the 50 drugs that were most heavily advertised to consumers accounted for almost half the $20.8 billion increase in drug spending last year...The remainder of the spending increase came from 9850 prescription medicines that companies did not advertise or advertised very little...Only the United States and New Zealand permit advertising of prescription medicines to consumers.“
Petersen M. New York Times. November 21, 2001
Noted by JFL, MD
This study was supported by the American Academy of Pediatrics and the Future of Pediatric Education II Project. Dr Pan was funded by a Primary Care Research National Research Service Award from the Health Resource and Services Administration.
We acknowledge the contributions of the American Academy of Pediatrics Resident Section and the American Academy of Pediatrics Committee on Pediatric Workforce.
- Received May 8, 2001.
- Accepted September 6, 2001.
- Reprint requests to (R.J.P.) Department of Pediatrics, University of California Davis Medical Center, 2516 Stockton Blvd, Rm 335, Sacramento, CA 95817. E-mail:
The views expressed in this article are solely those of the authors.
Dr Brotherton was formerly with the American Academy of Pediatrics at the time of her participation in the study; however, she now is with the American Medical Association.
- ↵American Medical Association. Physician Characteristics and Distribution in the US, 2001–2002 Edition. Chicago, IL: American Medical Association;2001
- Brotherton SE, Tang SS, O’Connor KG. Trends in practice characteristics: analyses of 19 periodic surveys (1987–1992) of Fellows of the American Academy of Pediatrics. Pediatrics.1997;100 :8– 18
- ↵Stoddard JJ, Back MR, Brotherton SE. The respective racial and ethnic diversity of US pediatricians and American children. Pediatrics.2000;105 :27– 31
- ↵Brotherton SE. Pediatric subspecialty training, certification, and practice: who’s doing what? Pediatrics.1994;94 :83– 89
- ↵Brotherton SE. Career plans of new pediatricians: results from a survey of residency program directors. Pediatrics.1991;88 :861– 866
- ↵American Academy of Pediatrics. Physician Workforce Ratios for Child Health, 1998. Elk Grove Village, IL: American Academy of Pediatrics;2000
- Copyright © 2002 by the American Academy of Pediatrics