Embedded within the current controversy over resident work hours are concerns that a proposed reduction in work hours would merely be offset by an increase in moonlighting, resulting in no significant change in the pattern of chronic sleep deprivation and fatigue experienced by most residents.1 To examine these concerns and to update the data on self-reported moonlighting and indebtedness by residents, the authors report findings from a large national random survey of PGY1 and PGY2 residents during the 1998–99 training year.
Despite the long work hours and sleep deprivation of residency and the active discouragement of some medical educators and administrators, physicians-in-training have long engaged in “moonlighting,” or work outside their programs.2,3 In 1978–79, 40.6% of residents at all levels who responded to a national random survey stated that they had moonlighted during the year.4 In 1982–83, a similar survey revealed this figure to be 32.8%, a 7.8% decline.4 In the latter survey, there was a significant variation by year of training, with only 7.5% of all PGY1 residents reporting moonlighting, compared with 38.4% of PGY2 residents and 53.4% of PGY3 residents.5 In a 1987 survey, these figures remained relatively constant, at 7.1%, 39.5%, and 53.3%, respectively.5
In 2000, a national survey of residents in obstetrics and gynecology by Defoe and colleagues revealed that only 12.7% of the 4,583 respondents at all levels of training were currently moonlighting.1 Certain other specialties, such as emergency medicine, however, continue to report figures near 50%.6 As in the earlier surveys, Defoe et al. found a marked variation by year of training, with 0.5% of PGY1 and 24.3% of PGY4 residents reporting moon-lighting.1 In addition to level of training, differences have been noted by specialty, gender and marital status.5
In the 1983 and 1987 surveys referenced above, the average hours of moonlighting reported by all residents were 14.3 and 13.4 hours per week, respectively, and mean figures for annual income derived from moonlighting were $7,400 in 1983 and $7,100 in 1987.5 Reported mean income for emergency medicine residents who moonlighted in 1996–97 was $25,000.6
In 1975, 56% of graduating medical students entered residency with an average educational debt of $12,400.4 By 1978, these figures were 65% and $14,300, and in 1982, they had reached 80% and $21,700, a rise of over 50% from 1978.4 Among 1997 graduates, 83.2% had educational debt, with 46% of these reporting levels of $75,000 or higher.7 Although compensation for residents has increased over the years, salaries have not kept pace with the consumer price index.4 Average salaries were $19,700 in 1983 and $20,200 in 1987. Indebtedness, meanwhile, rose from $17,200 in 1983 to $23,700 in 1987.5
In April 1999, the authors conducted a national random survey of PGY1 and PGY2 residents regarding their learning and work environments during their current residency year (1998–99). A 14.8% sample of first- and second-year residents (n = 6,016) was selected, using the Graduate Medical Education Database, which is secured as part of the annual survey of graduate medical programs by the American Medical Association.8 Residents from Puerto Rico and very small specialties such as nuclear medicine and medical genetics were excluded. Apart from this, there was no attempt at stratification. Deaths, early departures, residents in the wrong training year, and undeliverable surveys accounted for deletions from the sample, leaving 5,616 survey recipients.
A five-page questionnaire containing a number of questions about a variety of learning and work-related experiences was mailed to the identified residents together with a letter of explanation, a stamped, self-addressed return envelope, and a postcard on which the residents could indicate the separate return of the survey or their lack of willingness to participate. This process enabled the resident to remain anonymous. Residents who did not return the postcard were sent three follow-up mailings. In addition, program directors were mailed a letter asking them to encourage resident participation.
Included in the survey were the questions forming the primary data for this report. The first asked, “At graduation from medical school, what was your total educational loan indebtedness from college and medical school?” A space was provided for the “approximate amount.” Another question inquired, “During your current year of residency, did you work outside of your program to earn money?” (Yes or No). A follow-up question stated, “If yes, how many hours per month, on average, did you work?”
The survey instrument was pilot tested for clarity, relevance, and appropriateness, and cognitive interviews were conducted with a number of residents in different specialties. Revisions were made based on their feedback. Means, medians, and standard deviations were calculated for a number of relevant variables, including gender, marital status, level of training, and medical specialty. Data were analyzed by means of t-tests, ANOVAs, or Pearson correlations, where appropriate.
Of the 5,616 residents surveyed, 3,604 responded (64.2%). The response rate for PGY1 residents was 59% (n = 1,665) and that for PGY2 residents 68.7% (n = 1,912). Twenty-seven residents failed to indicate their postgraduate years. Response rates were consistent across the specialties, with only two surgical specialties registering below 50%. Distribution of residents across specialties approximated the national distribution with the exception of the two surgical specialties mentioned above. There were 2,016 men (56.5%) and 1,560 women (43.5%). U.S. medical graduates (USMGs) comprised 72.3% (n = 2,590) of the respondents, while 23.7% (n = 849) were international medical graduates (IMGs), and 3.4% (n = 132) were graduates of U.S. osteopathic schools.
Only 2.3% of PGY1 residents (n = 39) reported moonlighting during the 1998–99 year, averaging 14.9 hours per month each. All were married with children. For PGY2 residents, these figures were significantly larger, with 304 residents (15.9%) reporting an average of 23.1 hours per month each. Because so few PGY1 residents reported moonlighting, this report is limited to data from the PGY2 residents.
There were significant differences in moonlighting by specialty, with only 2.8% of PGY2 general surgery residents involved in such activity, as compared with 33.2% of residents in family practice, 28.7% in emergency medicine, and 23.8% in orthopedic surgery. The percentages of PGY2 moonlighters in each specialty are presented in Table 1, as are the average hours per month for each resident. It is of note that the variations in hours of moonlighting recorded for the different specialties have relatively little relationship to the percentages of residents actually moonlighting. For example, while 23.8% of orthopedic residents reported an average of 18.2 hours of moonlighting per month, the 17.4% of psychiatry residents who reported moonlighting averaged 29.5 hours per month.
Levels of reported indebtedness among the PGY2 residents varied widely, from zero to a high of $280,000 (by a non-moonlighter!). The overall average debt level was $59,232. For USMGs, it was $72,929, while for IMGs, the average was $12,771. Nearly a fourth of the USMGs reported debts of over $100,000, compared with only 4% for IMGs. Complete absence of indebtedness was reported by 13.6% of USMGs, while the figure for IMGs was 76%. For USMGs (n = 1,124) who reported debt over $10,000, the average was $86,767, while the corresponding figure for IMGs (n = 76) was $71,293. There was no significant difference in indebtedness between men and women ($72,294 vs. $73,647). Married residents ($70,543) reported less debt than did single residents ($75,696), and those with children had less debt ($70,519) than did those without children ($73,647). Divorced residents had the highest debt levels ($82,068).
For USMGs in their PGY2 year (Table 1) there were significant differences in indebtedness by specialty, ranging from a low of $32,250 for residents in radiation oncology to a high of $107,900 for those in physical medicine and rehabilitation. Residents in general surgery ($76,470), family practice ($74,603), pediatrics ($73,982), and internal medicine ($70,317) were all in the mid-range. The average indebtedness for USMGs who did not moonlight was $71,691 and for those who did, it was $77,867. The same figures for IMGs were $11,970 and $26,384, respectively.
Variations by Types of Residency Experience
As illuminating as are the data by specialty, it seems useful to examine both moonlighting and indebtedness by the four types of residency experience we have described elsewhere: high intensity, moderate intensity, low intensity, and moonlighters (see Table 2).9 These distinct residency types, derived by SPSS K-cluster analysis, provide a typology of residency experience, using a number of demographic and work-related variables, that appears to override the limitations inherent in a strict focus on specialty, such as gender and school of graduation.
Viewed in this framework, rather than being more stressed, the moonlighting group reported significantly less stress, fewer work hours, more sleep, and higher satisfaction than the high-intensity group and were most like the low-intensity group on nearly every measure. In addition, they were mostly men, USMGs, PGY2s, married, and with children. Moonlighters reported significantly less stress (4.49 vs. 4.83) than did non-moonlighters, and were less likely (1.5% vs. 3.4%) to have been named in a malpractice suit (p < .0001). In addition, for the PGY2 residents in this report, there were significant differences in the numbers and percentages of moonlighters in the four groups, as well as in the numbers of moonlighting hours per month (p < .001). Most important, there were no significant differences in the debt levels between the moonlighters and the non-moonlighters for any of the residency types.
This report has confirmed the recent trend towards an overall decrease in moonlighting on the part of residents, especially in their first year of training, along with continuing increases in reported indebtedness.1,4,5,7,10 The data suggest that while levels of indebtedness undoubtedly play an important role in the decision to moonlight, this is not the sole determinant. For example, licensure requirements generally exclude most PGY1 and IMG residents. In addition, moonlighters appear to be a unique group of residents with distinct characteristics, whose actions involve a combination of other factors, including available time, needed skills, and motivation.
First, there must be time. In our study, except for orthopedic surgery, most moonlighters came from specialties with only moderate numbers of work hours. Indeed, selecting a specialty and a residency program with fewer demands may even be a deliberate strategy for some moonlighters. Shift limits, such as those in emergency medicine, also make it possible to schedule predictable time for moonlighting. Some programs even offer in-hospital moonlighting opportunities.11 Second, one must have the required skills. Again, emergency medicine and orthopedic residents appear to be singularly equipped to function easily and effectively in emergency rooms, surgicenters, and ambulatory clinics.
Finally, motivation is essential. High levels of debt undoubtedly create pressure on the resident to look for ways to ease the financial burden, especially if a divorce or a spouse and children are involved. However, there is evidence that medical students, at least, appear to be taking on additional debt rather than cutting back on their standards of living.10 Finally, moonlighters may secondarily be motivated by a desire for more independence and autonomy, as well as for greater experience and learning. One large study of emergency medicine residents found that 94% of both moonlighters and non-moonlighters “believed that moonlighting enhanced residency academic performance” and that “the most frequently cited motivation for moonlighting among moonlighting residents was the educational experience it afforded.”6p.148
The limitations of the present study are those found with all self-report surveys. Short of direct observation and verification, there is no way to be sure that such responses are accurate and not self-serving. However, this study reached a large, national, random sample with a good response rate, and the results are consistent with studies by other investigators.1,4,5 In addition, the figures reported here are averages and do not adequately reflect individual variations. information about PGY3 and later residency years also would have been informative, but reaching such residents at the end of their third or later years is extremely difficult. Our overall aim was to better understand the breadth of the basic core residency experience.