Gynecologists are increasingly a source of general preventive health services for women in the United States,1 often providing cancer screening and other preventive health care for women at higher rates than women under the care of general medical physician primary care providers alone.1–4 Recognizing this trend, managed care plans have granted women open access to gynecologists without gatekeeper referrals, which in turn, has opened the way for gynecologists to take a more prominent role in the management of women’s primary care.
The appropriateness of this “open access” has been examined in a variety of studies in past years, but important methodological limitations have raised questions about the generalizability of their findings. These limitations include small sample sizes,3 small geographic areas,4 and sampling bias.2 Further, prior studies that directly compare performance of gynecologists and general medical physicians have not accounted for the potential effect of selection bias—the association of a woman’s preference for gynecologists over general medical physicians, or vice versa, with their preference for receiving preventive health care—on different rates of preventive health care use. Lastly, previously published studies have generally not examined the impact of receiving care from both gynecologists and general medical physicians, an important consideration given that most women fall into this category in previous studies. Such a pattern of care could lead to higher rates of preventive health care or even duplication of care.
The aim of the current study was to estimate how preventive services and counseling differ for women seen by general medical physicians and gynecologists and whether seeing both types of physicians had a greater impact on delivery of gender-specific and gender-neutral preventive care than by either type of physician alone. Based on previous research of collaborative (generalist-specialist) health care,5 we hypothesized that women seeing both general medical physicians and gynecologists would report receiving more preventive services and counseling than women seen by either type of physician alone. Our analyses used data from a large, nationally representative survey and involved multivariable regression and propensity score stratified analyses to adjust for the possibility of selection bias associated with generalist and gynecologist care.
MATERIALS AND METHODS
We used data from the 2000 National Health Interview Survey (NHIS). The National Health Interview Survey is an annual, cross-sectional, and nationally representative survey of the civilian, noninstitutionalized population of the United States administered by the National Center for Health Statistics.6 The National Health Interview Survey uses a multistage area probability sampling design that involves interviews of households from 678 primary sampling units in all 50 States and the District of Columbia. A primary sampling unit consists of a county, a small group of contiguous counties, or a metropolitan statistical area. A single adult (age 18 years and older) within each household is interviewed in depth about her own sociodemographic characteristics, health status, and health care services use. We used data from the year 2000 administration of the National Health Interview Survey because it included a Cancer Control Supplement that asked specific questions about providers seen and preventive services received. The 2000 survey was conducted in person, in English or Spanish, and involved 32,374 persons. The response rate was 82.6%. Because this study only used publicly available, deidentified data, it was exempt from review by the Mount Sinai School of Medicine Institutional Review Board.
Our study focused on women ages 18–64 years who could be reasonably cared for by general medical physicians, gynecologists, or both. Hence, our analyses excluded individuals with histories of cervical, breast, uterine, and ovarian cancers; pregnancy at the time of the interview; and those receiving Medicare coverage because such individuals are more likely to have major disabilities. For analyses of cervical cancer screening (Pap tests), we additionally excluded women who underwent hysterectomy. We further limited our sample to women under the age of 65 to use unambiguous screening guidelines to define our outcomes, and we limited our study sample to women who had at least one visit to a gynecologist or generalist (internist or family practitioner) in the past year as a proxy for access to care.
We examined seven age-specific preventive care services and counseling topics that are recommended by the U.S. Preventive Services Task Force7 and were captured in the 2000 National Health Interview Survey. We studied four preventive services outcomes for women aged 18–64 years: 1) cervical cancer screening within 3 years (Pap tests), 2) diet counseling, 3) exercise counseling, and 4) tobacco use screening. For women aged 50–64 years, we studied receipt of the above four preventive services, and three additional age-specific preventive services: 1) mammography within 2 years, 2) clinical breast examination within 2 years, and 3) colorectal cancer screening. Individuals were considered to have had colorectal cancer screening if they underwent any of the following screening tests: fecal occult blood testing (office or home) within the past 2 years, sigmoidoscopy in the past 5 years, or colonoscopy in the past 10 years. Individuals with a history of colorectal cancer were excluded from the analysis of colorectal cancer screening. We further categorized the preventive services as gender-specific (Pap test, mammography, clinical breast examination) and gender-neutral (diet and exercise counseling and screening for tobacco use and colorectal cancer).
The independent variable of interest was the specialty of the physician(s) from whom the study respondent received care. Respondents to the 2000 National Health Interview Survey were asked to identify which type of physicians they visited in the prior 12 months from a list of physician specialties that included “gynecologist” and “internal medicine or family practice physician.” Hence, we defined the physician specialty variable as gynecologist alone, general medical physician alone (internist or family practitioner), or both gynecologist and general medical physician. Our analyses included additional independent variables associated with receipt of appropriate preventive services in women who also were available to us in the National Health Interview Survey data set.8–16 These variables included age, race, education, income, insurance, census region, urban residence, and immigration status. Race was dichotomized into white non-Hispanic and all other groups in adjusted analyses. We analyzed measures of health status including self-reported health status, hospitalization within the past year, number of physician visits in the past year, and smoking. Because management of complicated diseases might restrict a physician’s ability to provide comprehensive preventive services, we also controlled for comorbid illness using a composite measure for the presence or risk of cardiovascular diseases (hypertension, diabetes, cerebral vascular disease, and cardiac disease), a composite measure for pulmonary disease (asthma or chronic obstructive pulmonary disease), and depression. Finally, additional gynecologic variables that might affect the use of Pap tests or mammography were analyzed, including current and previous use of oral contraceptives and history of abnormal Pap tests or abnormal mammograms.
We examined the association of provider type (gynecologist, generalist, or both) with receipt of the seven preventive services, using a weighted χ2 test. We then used logistic regression to model the association of provider type with each outcome, controlling for other independent variables. Models were constructed using a manual stepwise procedure (entry and stay criteria, P<.20), and the adjusted odds ratios were converted to relative risks.17 In addition, we examined the association of provider type with the total count of preventive services received using Poisson regression.18 The maximum possible number of services was four for women aged 18–64 years and seven for women aged 50–64 years. We repeated all of these analyses in a subgroup of women who lacked major comorbidities (cardiovascular disease, hypertension, diabetes, and pulmonary diseases) because the presence of such chronic illnesses would predispose them to seek care from a general internist.
A woman’s choice of health care provider (gynecologist or generalist) could also be associated with how her preventive care is prioritized. For example, a patient with many medical problems might opt to receive all her care from a generalist and might not receive all preventive services because of the competing interests of her chronic illnesses. To mitigate the effects of this potential selection bias, we conducted a set of propensity score analyses, stratifying women by their propensity to see a generalist.19 Initial analyses showed that women who had seen gynecologists only had similar characteristics and rates of preventive services outcomes as women who had seen both gynecologists and generalist medical physicians. We, therefore, combined these two provider groups in the propensity score analysis. To calculate propensity scores, we fit a logistic regression model estimating subjects’ propensity to see a gynecologist or a gynecologist and a generalist (as opposed to a generalist alone). The independent variables in the model included all sociodemographic and health status variables shown in Table 1. The sample was then stratified by quintiles of propensity score. Within each quintile we examined the balance of covariates between women who saw gynecologists and general medical physicians and those who saw general medical physicians alone using the χ2 test. Finally, within each quintile of propensity score, we used the χ2 test to determine the association between physician specialty and preventive health services received.19 All analyses used the 2000 National Health Interview Survey sampling stratum variables, primary sampling unit variables, and survey weights as suggested by the National Center for Health Statistics.20 The sampling weights provided by the National Health Interview Survey were based on sampling design, nonresponse, and poststratification adjustments.20 The statistical software SUDAAN 8.2 (Research Triangle Institute, Cary, NC) was used for all analyses to account for the clustered, hierarchical sampling design of the National Health Interview Survey.
Of the 32,374 people included in the overall National Health Interview Survey, 7,317 women met inclusion criteria, representing 43.6 million women nationally; 5,766 (78.8%) were 18–49 years of age and 1,551 (21.2%) were 50–64 years of age. Table 1 shows the characteristics of the sample stratified by the type of physician(s) they saw in the prior 12 months. Women who visited gynecologists alone or gynecologists and general medical physicians were younger and tended to have markers of higher socioeconomic status than women who saw general medical physicians alone, including higher levels of education, higher incomes, and private insurance. They also tended to have better health, as indicated by better self-rated general health and lower prevalence of chronic diseases (Table 1).
The distribution of preventive services and types of counseling received varied widely (Table 2). Overall, 88–95% of women were up to date with breast and cervical cancer screening, whereas only one quarter had ever had a discussion about diet or exercise with their doctor or had undergone colorectal cancer screening (among those 50–64 years of age). Among women aged 18–49 years, the average total number of services was 1.6 for survey respondents seen by a general medical physician alone, 2.2 for those seen by a gynecologist alone, and 2.4 for those seen by both types of physicians (P<.001). Among women aged 50–64 years, the average number of services were 3.4 (general medical physician), 3.7 (gynecologist), and 3.5 (both) (P<.003).
For all of the services examined, women seeing gynecologists alone reported receiving the preventive care as frequently as or more frequently than women seeing only general medical physicians (Table 2). Significant differences across the provider groups were observed mostly for the gender-specific preventive services; women seeing gynecologists and those seeing gynecologists as well as general medical physicians were significantly more likely to have had Pap tests, mammograms, and clinical breast exams. Among the gender-neutral preventive services, tobacco use screening occurred more frequently among woman receiving some care from gynecologists, whereas there were no differences across provider groups for the other gender-neutral services. Comparing care between the gynecologist only and general medical physician plus gynecologist provider groups, the combination of provider types only significantly increased the rate of exercise counseling (P=.04).
Across the three provider groups, statistically significant differences remained after adjusting for demographic and health status (Table 3). Notably, in the multivariable analysis, the combination of general medical physician and gynecologist care increased the probability that women would receive diet and exercise counseling and colorectal cancer screening. Quantitatively similar results were observed in a subgroup analysis of women who lacked cardiovascular disease, cardiovascular risk factors, or pulmonary diseases (data not shown).
In a Poisson regression analysis, the total count of services received by women aged 18–49 years remained significantly greater among those seeing gynecologists (P<.001) or gynecologists plus general medical physicians (P=.001) compared with women seeing general medical physicians alone, even after adjusting for age, race, education, income, insurance, and health status. However, significant differences were not observed for women 50–64 years old (P=.07 and P=.40, respectively).
In the propensity score analysis, we compared women seen by general medical physicians alone with those seen by gynecologists alone or both gynecologists and general medical physicians (n=6,139). Across the five strata of propensity scores, there were no significant differences between the women seen by general medical physicians and those seen by gynecologists alone or gynecologists plus general medical physicians for any of the demographic or health status variables, indicating that the propensity score adjustment effectively balanced the two groups on observable variables (data available on request).
The propensity score analyses were largely consistent with the findings of the nonpropensity score adjusted multivariable logistic regressions (Table 4). Across all five quintiles, women who saw gynecologists were more likely to have had a Pap test, and no differences were observed in the rates of diet and exercise counseling and colorectal cancer screening. Patterns of mammography and clinical breast examination were generally consistent across the five quintiles, although statistical significance was not achieved in all quintiles. The sole exception was for tobacco use screening, which in contrast to the simple logistic regression, did not differ between provider groups.
In this study, we found that women who received care from a gynecologist, alone or in combination with a general medical physician, were more likely to receive preventive health services, such as cancer screening and healthy behavior counseling, than women who saw general medical physicians alone. As predicted, the largest differences were observed for gender-specific care, such as breast and cervical cancer screening. Women under the care of gynecologists were also more likely to receive tobacco cessation counseling. Notably, women under the exclusive care of gynecologists were as likely to have counseling for diet and exercise and to undergo colorectal cancer screening as women who received care from general medical physicians. These findings demonstrate that gynecologists perform as well as or better than general medical physicians in the delivery of the seven preventive services we studied. From the standpoint of preventive health care, these findings validate policy decisions that facilitate women’s access to gynecologists, such as identifying gynecologists as primary care providers in managed care plans. They also add credibility to women’s decisions to rely on gynecologists to receive basic health care screenings and counseling. It is important to note, however, that patients of both gynecologists and general medical physicians had overall low rates of gender-neutral preventive care: overall, only about half of women in the sample had colorectal cancer and tobacco use screening and about one quarter had diet and exercise counseling. As such, comprehensive preventive health care remains suboptimal for women seeing either type of provider.
Our findings are consistent with previous studies that address delivery of primary preventive care to women,1,3,4 although some important differences exist. First, we found significantly lower rates of visits to both general medical physicians and gynecologists than the rates described in previous studies. In both the Henderson et al study4 and the Kaiser Women’s Health survey,15 more than 50% of the women reported seeing two providers. These studies included visits to physicians at any time in the past, whereas we focused on visits during the preceding 12 months. This methodological difference may account for the observed discrepancy in rates of visits to both general medical physicians and gynecologists. Secondly, similar to findings in previous studies, there was a high rate of screening mammograms, clinical breast exams, and Pap tests. The low rate of screening for colorectal cancer is comparable to previous research.12
The effect of provider type on rates of screening in a large national sample is unique to this study. In this study, we show that the increase in services received among the women seen by both may be equivalent to that received by women seen by gynecologists only. Women who saw gynecologists alone or in conjunction with a general medical physician received more services than women seen only by general medical physicians. As opposed to previous research, we did not find a difference in counseling between the different provider types.1,3,4
As with many studies that use publicly available data for secondary data analyses, interpretation of our findings should be viewed in light of the data’s limitations. First, we were unable to examine use of other important preventive health care services, such as cholesterol and blood pressure screening. Second, use of health care services documented in the National Health Interview Survey was assessed through self-report and may be prone to recall bias. Prior research has consistently shown that cervical cancer screening and mammography are overestimated by self-reports.11,21,22 If women confuse simple speculum exams with cervical cancer screening, then overestimates of cervical cancer screening might be greater among women receiving care from gynecologists than those seeing general medical physicians alone. Third, the National Health Interview Survey includes only 2 years of data for individuals, and some screening measures are not necessarily indicated on an annual basis. Thus, some individuals included in our analyses may have had screening within a guideline-recommended time frame that was not captured in the National Health Interview Survey data. Additionally, study subjects reported only care received from general medical physicians and gynecologists in the past 12 months. Because some women may have received care from either type of physician before the look-back period, it is possible that some preventive services attributable to care received during that time were associated in our analyses with a physician in the period of observation. Bias resulting from this methodological limitation might tend to overestimate the association between gynecologist care and use of preventive services since women, particularly younger and healthier ones, may be more likely to make annual visits to gynecologists than to general medical physicians. However, our findings reveal strong associations between gynecologist care and preventive care when both were received in the past 12 months, suggesting that this may not be an important bias. Moreover, data on visits to general medical physicians and gynecologists was also self-reported and may be susceptible to recall or identification bias, although data from prior research are lacking on this topic.
Our analysis included no information about physician gender. Female physicians have been shown to order more women’s health-related preventive services,16 and gender distributions may have differed among gynecologists and general medical physicians in our study.
Finally, our analyses focused on seven preventive screening and counseling measures. Our findings therefore cannot be extrapolated to nonpreventive health care.
In conclusion, we found that women who received care from both gynecologists and general medical physicians were more likely to receive guideline-recommended preventive services than women who received their care from general medical physicians alone. Our study also found that women who saw gynecologists alone were more likely to get gender-specific services with similar rates of gender-neutral preventive care. Yet, many women, even those seen by both gynecologists and general medical physicians, received no or few preventive health care services. Although our findings support current policies that provide open access to gynecologists for women in managed care and other health insurance plans and validate the role of gynecologists as providers of general preventive health care for women, both gynecologists and general medical physicians must remain vigilant about the prevention of diseases in women for which good screening measures exist.
1. Scholle SH, Chang JC, Harman J, McNeil M. Trends in women’s health services by type of physician seen: data from the 1985 and 1997–98 NAMCS. Womens Health Issues 2002;12:165–77.
2. Wallace AE, MacKenzie TA, Weeks WB. Women’s primary care providers and breast cancer screening: who’s following the guidelines? Am J Obstet Gynecol 2006;194:744–8.
3. Gallagher TC, Geling O, Comite F. Use of multiple providers for regular care and women’s receipt of hormone replacement therapy counseling. Med Care 2001;39:1086–96.
4. Henderson JT, Weisman CS, Grason H. Are two doctors better than one? Women’s physician use and appropriate care. Womens Health Issues 2002;12:138–49.
5. Ayanian JZ, Landrum MB, Guadagnoli E, Gaccione P. Specialty of ambulatory care physicians and mortality among elderly patients after myocardial infarction. N Engl J Med 2002;347:1678–86.
8. Phillips KA, Morrison KR, Andersen R, Aday LA. Understanding the context of healthcare utilization: assessing environmental and provider-related variables in the behavioral model of utilization. Health Serv Res 1998;33:571–96.
9. Mehrotra A, Zaslavsky AM, Ayanian JZ. Preventive health examinations and preventive gynecological examinations in the United States. Arch Intern Med 2007;167:1876–83.
10. Fiscella K, Franks P, Doescher MP, Saver BG. Disparities in health care by race, ethnicity, and language among the insured: findings from a national sample. Med Care 2002;40:52–9.
11. Fiscella K, Holt K, Meldrum S, Franks P. Disparities in preventive procedures: comparisons of self-report and Medicare claims data. BMC Health Serv Res 2006;6:122.
12. Kiefe CI, McKay SV, Halevy A, Brody BA. Is cost a barrier to screening mammography for low-income women receiving Medicare benefits? A randomized trial. Arch Intern Med 1994;154:1217–24.
13. O’Malley AS, Forrest CB, Mandelblatt J. Adherence of low-income women to cancer screening recommendations. J Gen Intern Med 2002;17:144–54.
14. Sambamoorthi U, McAlpine DD. Racial, ethnic, socioeconomic, and access disparities in the use of preventive services among women. Prev Med 2003;37:475–84.
16. Cassard SD, Weisman CS, Plichta SB, Johnson TL. Physician gender and women’s preventive services. J Womens Health 1997;6:199–207.
17. Zhang JX, Iwashyna TJ, Christakis NA. The performance of different lookback periods and sources of information for Charlson comorbidity adjustment in Medicare claims. Med Care 1999;37:1128–39.
18. Neter J, Kutner MH, Nachtsheim CJ, Wasserman W. Applied linear regression models. 3rd ed. Chicago (IL): Richard D. Irwin, Inc.; 1996.
19. D’Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265–81.
21. Bowman JA, Sanson-Fisher R, Redman S. The accuracy of self-reported Pap smear utilisation. Soc Sci Med 1997;44:969–76.
22. Holt K, Franks P, Meldrum S, Fiscella K. Mammography self-report and mammography claims: racial, ethnic, and socioeconomic discrepancies among elderly women. Med Care 2006;44:513–8.
Figure. No caption available.