Our study provides the most recent data showing that national rates of cervical cancer screening are continuing to decline in older women and women who have undergone a hysterectomy. Over the time period we examined (1993 to 2005), NHIS data showed consistent patterns of lower Pap test use among older women (65+) compared to younger women regardless of hysterectomy status. Women who reported a hysterectomy were significantly less likely than other women in their age group to have a recent Pap. Our findings also document for the first time in a national dataset significant associations between measures of health (as proxies for life expectancy) and use of Pap tests among older women.
The data suggest that screening rates are lower for older women in poorer health (e.g., 57.2% with a recent Pap among women reporting excellent/good health compared to 52.5% of women in fair/poor health,
0.03). Still, over half of women reporting poor health, a chronic disability, or a CCI of 4 or greater reported a recent Pap test. This finding indicates overutilization of screening among women who are unlikely to benefit from early detection. Our results are similar to findings from a 2000–2001 study conducted in California that showed overutilization of breast and cervical cancer screening among older women in poor health.1
It is unclear why women who seemingly won’t benefit from continued Pap testing still get screened. Some investigators have suggested that older women in poor health may be more motivated to get screened because of a greater sense of susceptibility to illness. They also may have more opportunities to screen because of frequent interactions with healthcare providers.31
Another possibility is that decisions to discontinue screening would involve a conversation between clinicians and patients about predicting life expectancy and there are no specific clinical guidelines on how to do this. Further, older adults may not understand concepts of uncertainty and probability that underlie understanding how competing causes of mortality and delayed benefits of screening influence the prudence of continued screening.32
A national survey of women 40 and older found that women were resistant to reducing the frequency of Pap tests as recommended by revised clinical guidelines especially if they thought a reduction was being suggested for cost rather than scientific reasons.33
Similarly, a qualitative study of attitudes about continuing cancer screening found that 43% of older adults would consider screening even if their doctor recommended against it.32
It is not clear whether these participants understood that the benefits of screening may not be realized for years or whether having this knowledge would make a difference in their enthusiasm for screening. Although patient preferences are important to consider, there is some evidence that physicians also account for patient health status in deciding whether to test,34
and our data support this.
Although recent Pap use was lower among all women who reported hysterectomies during the time period we examined (1993–2005), about half of older women with hysterectomies were screened within the three years prior to interview in 2005. Since we excluded women with a history of cervical cancer from our analysis, the women with a hysterectomy in our sample likely had the procedure for benign disease, suggesting that Pap testing was unnecessary. Our findings are not inconsistent with Behavioral Risk Factor Surveillance System data from 1999 to 2002 that showed more than two-thirds of hysterectomized women reported a recent Pap (a higher rate of inappropriate screening compared to our data).35
More of a decline might have been expected in our 2005 study both because the USPSTF recommended in 1996 that Pap testing is unnecessary for women who had a hysterectomy with removal of the cervix for benign disease and because of the considerable publicity generated by the earlier study.
We examined three different health indicators. Both poorer general health and chronic disability were independently associated with declining Pap use. Although the CCI did not reach statistical significance in our multivariate analysis, the pattern of association was consistent with the other two measures. Our findings for general health status are consistent with Walter and colleagues.1
With regards to chronic disability, research has shown that women with major lower extremity mobility difficulties are less likely to receive Pap tests.36
We could not ascertain this distinction with the NHIS data.
For a woman with short life expectancy, it would be clinically appropriate to forego cancer screening. Thus, the lower rate of screening among women with poor health status in our sample most likely represents a responsible population pattern. However, the lower screening rate among women with disabilities should be viewed with more caution. In our analysis, we could not distinguish whether the specific disability would reduce life expectancy. If it would not, then cancer screening would still be appropriate. Clinicians and health care systems need to facilitate access to cancer screening among disabled women with good life expectancy. The lack of a significant association between the CCI and screening may mean that clinicians do not pay attention to life expectancy or medical disease burden when deciding whether to perform cervical cancer screening, or the CCI, which was originally created to predict mortality, may not be as sensitive in detecting an association with health care utilization as the other two health measures. An earlier study using NHIS found that the CCI was significantly associated with PSA testing among men aged 45–74 (1.09, CI=1–1.18,
.041) but not among older men (75+).37
It should be noted that, in this analysis, the CCI was adapted to the content of the NHIS and has not been previously validated with this particular survey instrument; however, the adapted index used here was a fairly close approximation of other claims-based applications. 38
Although we used data from a large, nationally representative sample with high response rates to assess the association between age, health and hysterectomy status and cervical cancer screening, there are some limitations. Because the NHIS is cross-sectional, we cannot ascertain causality. Thus we cannot conclude that health status is the reason older women are getting fewer Pap tests. Nor can we conclude that revised clinical guidelines recommending longer screening intervals led to discontinuing screening at older ages or caused the decline in Pap use that we observed between 2000 and 2005. Furthermore, self-reported data could overestimate adherence.39
We could not assess the proportion of hysterectomies that were supracervical or performed for cervical neoplasia with our data. Nevertheless, prior national estimates indicate that almost half of women who have undergone hysterectomy are receiving unnecessary Pap tests.35
Though age, health and hysterectomy status appear to influence Pap test use, current national data suggest that there still is inappropriate screening of older women. The behavioral model of health services utilization,14,15
as well as other health behavior models 40–42
have historically focused on the challenge of increasing delivery and uptake of recommended clinical practices. However, evidence of widespread overutilization here and elsewhere in health care,43,44
suggests that more theoretic research on the factors that influence and explain overuse of health services is needed. Models have been applied to the overuse of antibiotics 45
and diagnostic tests 46,47
and extending similar efforts to understand individual, provider and system factors that encourage overutilization of screening would benefit both patients and society by avoiding tests that produce unnecessary risks and costs.
Informed decision making approaches, such as decision aids, could potentially reduce unnecessary Pap testing. For example, decision aids for prostate cancer screening have resulted in decreased interest in and use of PSA testing among patients seeking routine care.48
Yet informed decision-making approaches would be incomplete without incorporating information regarding life expectancy, the relevance of competing disease risks, and the delayed medical benefit of screening. Sometimes the ascertainment of unnecessary testing will be clear-cut, as in the case of women with a prior hysterectomy, and informational interventions will be relatively straightforward. Yet in more difficult cases, such as among women with advancing age or competing illnesses, informed decision-making approaches that help patients and clinicians better communicate about how these complex issues relate to cervical cancer screening will be beneficial.