Table 3 reports the results of multiple logistic regression analyses to estimate the increased risk of 2 types of abnormal Pap results (at least, atypical squamous cells of undetermined significance and only HSIL) associated with each of the independent characteristics of interest. For each of the 2 outcomes, we report 2 multivariable logistic regression models, the first including all covariates and the second including only those covariates that were significant at the P < .05 level, by using backward elimination.
Table 4 compares rates of each of the 5 categories of Pap test results among women screened in the inpatient screening program and those of patients screened at the hospital outpatient sites during the 4-year time period (1999–2002). Odds ratios (ORs) and 95% confidence intervals (95% CIs) are used to identify statistically significant differences in rates between the 2 groups.
During the period January 1999 to December 2002, a total of 86,697 female patients were admitted to the Johns Hopkins Hospital. Of these, 22% were aged 21 years or younger. Of the women admitted, 1,482 were approached for cervical cancer screening. These women had indicated at the time of hospital admission that they were interested in taking advantage of the in-house cervical cancer-screening program. Some women were referred to the screening program by their admitting physicians. Of the women who requested screening, a total of 1,117 (75.4%) underwent Pap test screening during their hospital admission.
The age distribution of the women who underwent screening was more heavily concentrated among women aged 20–60 years (84%) compared with the overall hospital population. The proportion of uninsured/self-pay patients in the screened population was slightly higher (26%) than in the overall hospital population (21%). Of the women screened, 62% were black and 35% were white compared with 44% and 51%, respectively, in the overall hospital population. The average length of stay for all inpatients during this time was 6.2 days (data not shown).
Reasons that initially identified women were not screened included general illness; specific contraindication to the examination, such as hip or back surgery or menses; psychiatric conditions, such as mania or dementia; and patients changing their mind for reasons not specified.
Table 2 shows that approximately half of the women screened were admitted to the internal medicine service, and another 25% were admitted to psychiatry. Fourteen percent were known to be human immunodeficiency virus (HIV) positive, although only 5% were admitted to the acquired immunodeficiency syndrome (AIDS) service. A substantial proportion of women screened reported a history of a sexually transmitted disease (STD) (28%) or a history of an abnormal Pap test (14%), and 57% of women screened reported having had a Pap test within the previous 3 years.
Sixteen percent of the inpatient Pap tests performed identified abnormal cervical cells, with 3% of tests identifying HSIL. Bivariate distributions in Table 2 suggest differences in the distribution of Pap test outcome by year, patient age, ethnicity, service, HIV status, and history of STD and abnormal Pap, but no substantial variation by insurance status or reported proximity of last Pap. Table 2 identifies several characteristics of women more likely to have HSIL Pap results: aged 16–40 years, black ethnicity, both admission to the AIDS service and positive HIV status, and history of STD and of previous abnormal Pap.
In Table 3, the final models identify 3 characteristics associated with significant risk for any abnormal test and 3 characteristics specifically predictive of HSIL. The 15.5% of women screened whose results showed atypical squamous cells, LSIL, HSIL, or atypical glandular cells of undetermined significance were significantly more likely to be younger, with a reduction in risk of 2% with each year of age older than the median age of 42 (OR 0.98; 95% CI 0.96, 0.99). Chart documentation or patient report of positive HIV status increased a woman's risk of abnormal results more than 6-fold (OR 6.32; 95% CI 4.30, 9.27). In addition, if the woman or her medical record reported that she had a history of a previous abnormal Pap, her risk of abnormal results more than doubled (OR 2.47; 95% CI 1.61, 3.77). No other variable, including whether or not the woman was on the AIDS service, retained significance in the multivariable model.
The second final model identifies characteristics of women whose Pap result identified HSIL, the abnormality most predictive of risk for invasive cancer. Black women in this population had almost a 4-fold risk of this result, adjusted for other variables in the model (OR 3.72; 95% CI 1.27, 10.93). In addition, HIV-positive status was associated with a 4-fold risk in HSIL result (OR 4.11; 95% CI 2.00, 8.43). As in the model for all types of abnormal results, a previous abnormal test adds a 2.5-fold risk for HSIL results (OR 2.42; 95% CI 1.12, 5.24).
In contrast to the inpatient screening program rate of 15.5%, during this time period, the overall rate of abnormal Pap results in the outpatient clinics (n = 111,933) was 7% (OR 2.44; 95% CI 2.07, 2.82) (Table 4). Although rates for each of the 4 categories of abnormal test are significantly higher in the inpatient screening program than the outpatient data, the greatest difference is seen in the prevalence of HSIL results, with an almost 5-fold increase in the rate of high-grade lesions (3% versus 0.7%; OR 4.63; 95% CI 3.28, 6.52).
We evaluated the effectiveness of an inpatient, hospital-based cervical cancer screening program at a single institution and found that the prevalence of high-grade preinvasive disease, the immediate precursor to cervical cancer, was between 4- and 5-fold higher in the inpatient cohort compared with all outpatient screenings performed in the Hopkins system during the same time period. During that time period, the prevalence of HSIL in our inpatient screening program was 3%, whereas the prevalence in our outpatient screening clinics was 0.7%. Although specific demographic data on the outpatient screening population are not available for these analyses, these outpatient settings serve a wide range of patients in both urban and suburban Baltimore and are therefore more similar to the general U.S. population. The Centers for Disease Control and Prevention Breast and Cervical Cancer Screening Program has reported a national HSIL prevalence of 1.1%4
We found that black race independently increased the chance of an HSIL Pap test in this self-selected group of patients. In addition, in our study, women known to be HIV seropositive were more likely to have HSIL Pap tests.
The strengths of this analysis include data spanning a 4-year period, including more than 1,000 women who requested and underwent cervical cancer screening during inpatient hospitalization at a single institution, that was read by one cytopathology laboratory. We compared rates of abnormal Pap tests to those observed in our combined outpatient clinics, which were read by the same laboratory during the same time period. The outpatient sample included more than 100,000 Pap tests.
Weaknesses of our data set include incomplete information on the screening history of our cohorts and demographic data, including risk factors, on the outpatient cohort. We also lacked data on some risk factors in the inpatient cohort. Screening histories obtained were by patient report. Patient recall of Pap test screening is often less than accurate.5,6 This cohort is a self-selected one in that the inpatients who are reported here requested screening after being asked by and large at the time of their hospital admission. We do not know how representative they are of our overall hospital admissions. All women admitted were offered a Pap test; most of those who declined in-house screening at the time of admission did so because they had established care and care providers already. Finally, although our cohort of women was large, our analyses may have had limited statistical power because of the relatively rare outcome events we were examining. Therefore, we cannot rule out that with larger sample or case-control design additional variables would be statistically significant predictors of abnormal Pap test results.
Data describing prevalence of abnormal Pap tests in hospitalized patients is scarce. Klassen et al7 surveyed 37 acute-care Maryland hospitals about their Pap screening policies. In 31 of 37 hospitals, women were routinely offered a Pap test. In most hospitals, however, this was not accompanied by an evaluation of a patient's individual need for a Pap test, patient education about cervical cancer, or a physician recommendation to the patient about screening. Only 2 hospitals had a specific person designated to counsel women about Pap tests and to perform them. Therefore, most hospitals complied with the letter of the law but did not effectively increase screening among their highest-need patients. Earlier work found that hospitalization appeared to increase the likelihood of a recent Pap test among women aged 45–54 years with household income more than $20,000 a year but not among older or poorer women.8
Why are women in the United States still being diagnosed with cervical cancer? Despite the widespread availability of Pap tests, the American Cancer Society estimated that in 2003, 13,000 women will be diagnosed with cervical cancer and that 4,100 women will die of the disease (American Cancer Society, Facts and Figures 2003, www.cancer.org). The Pap test is an inexpensive, relatively noninvasive, socially acceptable screening test. In the developed world, there exists an extensive, although cumbersome and expensive, infrastructure for the evaluation and treatment of women with abnormal Pap tests. There is a low but appreciable false-negative rate, and compliance with follow-up of abnormal results can be problematic. Perhaps the most important reason, however, is that many women do not get screened regularly. At least half of women diagnosed with cervical cancer have not had a Pap test in the 3 years previous to their diagnosis.2,3 Nationally, women who are less likely to undergo regular screening include women who are older, obese, poor, and rural, and whose native language is not English.
Certain chronic medical conditions, such as HIV infection, may be associated with less adherence to care, especially routine health maintenance. This analysis does not address the role of comorbidity, such as substance abuse, in our inpatient cohort. In women whose health is precarious enough to require hospitalization, routine health care issues such as screening may well be a lower priority for both the patient and for her health care providers. This analysis found that younger women represented a higher-than-expected percentage of those with high-grade lesions. Older age has been consistently shown to be a risk factor for noncompliance with screening guidelines and for cervical disease.9 The constellation of risk factors that increase a young woman's need for hospitalization may also place her at increased risk for cervical dysplasia. Conversely, older women may think that they do not need or want screening for cervical cancer and so may decline screening.
Hospital-based screening has been advocated as one potential opportunity to reach women who are not undergoing regular outpatient screening. In the 1970s, several states, including Maryland, New York, Ohio, and Hawaii, passed laws mandating that women admitted to hospitals be offered a Pap test. In general, however, these laws did not provide funding for such services. The development of an effective hospital-based cervical cancer-screening program requires several critical elements. These include designation of a specific individual to counsel and perform Pap tests and to follow up on the results, as well as individualized evaluation of each patient's need for a Pap test, counseling about cervical cancer and screening, and support for screening from the patient's admitting physician. In addition, a designated physician with expertise in the appropriate management of abnormal Pap tests must coordinate the program and direct follow-up management plans as needed. In our system, the annual budget, including a 40% full-time equivalent nursing position and Pap test supplies and processing costs, is approximately $50,000. The budget is supported by hospital nursing and administrative funds.
Our data suggest that hospital-based screening can capture women at increased risk for cervical disease. In terms of feasibility, in-house screening is reasonable considering that the average length of stay was 6.2 days and the screening nurse made rounds 3 times a week. Hospital-based screening, therefore, appears to provide a relatively low-cost, targeted intervention to improve screening for cervical cancer in a cohort of women at higher risk for disease than those accessing routine outpatient screening clinics. It must be emphasized that the inpatient population at an inner-city academic institution reflects a cohort with more risk factors for cervical disease than the general population. Therefore, inpatient screening would be more cost-effective in this setting than in hospitals serving more heavily screened patients. Nonetheless, it appears that offering screening for cervical cancer at the time of hospital inpatient admission is a method of targeting a population that is at higher risk for precancerous lesions than the rest of the population and so represents an efficient use of limited screening resources.
1. Koss LG. The Papanicolaou test for cervical cancer detection. A triumph and a tragedy [review]. JAMA
2. Janerich DT, Hadjimichael O, Schwartz PE, Lowell DM, Meigs JW, Merino MJ, et al. The screening histories of women with invasive cervical cancer, Connecticut. Am J Public Health
3. Fruchter RG, Boyce J, Hunt M. Missed opportunities for early diagnosis of cancer of the cervix. Am J Public Health
4. Kurman RJ, Henson DE, Herbst AL, Noller KL, Schiffman MH. Interim guidelines for management of abnormal cervical cytology: The 1992 National Cancer Institute Workshop. JAMA
5. Newell S, Girgis A, Sanson-Fisher R, Ireland M. Accuracy of patients’ recall of Pap and cholesterol screening. Am J Public Health
6. Paskett ED, Tatum CM, Mack DW, Hoen H, Case LD, Velez R. Validation of self-reported breast and cervical cancer screening tests among low-income minority women. Cancer Epidemiol Biomarkers Prev
7. Klassen A, Hall A, Bowie J, Weisman CS. Improving cervical cancer screening in hospital settings. Prev Med
8. Klassen AC, Celentano DD, Weisman CS. Cervical cancer screening in hospitals: the efficacy of legislation in Maryland. Am J Public Health
© 2004 The American College of Obstetricians and Gynecologists
9. Sawaya GF, Sung HY, Kearney KA, Miller M, Kinney W, Hiatt RA, et al. Advancing age and cervical cancer screening and prognosis. J Am Geriatr Soc