Rates of Cervical Cancer Screening Among Women With Severe Mental Illness in the Public Health System

This retrospective cohort study included women between the ages of 18 and 67 who were enrolled in California Medicaid (Medi-Cal) and received care in a California specialty mental health clinic between October 1, 2010, and September 30, 2011. A detailed description of the cohort in the parent study has been published previously (11). Of the 31,308 patients included in the study, 38% were white, 19% were black, 19% were Hispanic, 13% were Asian or Pacific Islander, and 10% identified as another race-ethnicity. Most had schizophrenia spectrum disorders (42%) or major depressive disorder (29%). Most women (74%) had evidence of use of primary care in the prior year. [A table in an online supplement to this report presents these and other descriptive statistics.]

All data were extracted from Medi-Cal and the Client and Service Information system, which collects client-level data from California’s county mental health programs. The data set consisted of administrative, pharmacy, and billing information that was combined, deidentified, and provided to the study investigators by the Data and Research Committee of the California Department of Health Care Services. Rural counties with few observations were grouped with counties of similar size, region, and demographic characteristics. To estimate the annual cervical cancer screening rate of the general U.S. population, we solved the equation p₃=1–(1–p₁)³ for p₁, the annual rate, where p₃ is the 2010 estimate of the California triennial screening rate published by the Centers for Disease Control and Prevention (12). This estimator calculates the cumulative proportion of women screened within a three-year period as the complement of the proportion who remain unscreened throughout the three years, under the simplifying assumption that the annual screening rate is constant and does not depend on having been screened the previous year.

The primary outcome measure was evidence of cervical cancer screening from individual Medi-Cal records. This was identified through Current Procedural Terminology (CPT) codes regarding gynecologic cytology and Pap smear collection/screening (CPT/HCPCS 88141–88143, 88147, 88148, 88150, 88155, 88164, 88167, 88174, and 88175).

Potential predictors of cervical cancer screening included age, race-ethnicity, county of residence (rural versus urban), presence of a diagnosis of severe mental illness described in parent study (11), a history of drug or alcohol use as documented by mental health providers, and evidence of health care utilization. Severe mental illness was defined as a DSM-IV-TR diagnosis of a schizophrenia spectrum disorder, anxiety disorder, bipolar disorder, major depressive disorder, or another axis I diagnosis by a mental health provider.

Poisson models with robust standard errors were used to evaluate whether predictor variables were associated with cervical cancer screening. For each predictor, we fit two Poisson models. The first was unadjusted, and the second was adjusted for confounders identified by using a directed acyclic graph (or causal graph); any mediating effects were excluded. All data were analyzed with Stata Statistical Software, Release 13.2.

The study was approved by the University of California, San Francisco, Committee of Human Research (11–06939), the State of California Committee for the Protection of Human Subjects (FWA00000681), and the Data and Research Committee of the California Department of Health Care Services (IRB00000552).

Among Medi-Cal enrollees with severe mental illness between the ages of 18 and 67, 20.2% (N=6,332) received cervical cancer screening within the one-year study period. The percentage of women of various races and ethnicities who were screened for cervical cancer ranged from 19.0% among white women (N=2,242 of 11,809), 20.9% among black women (N=1,265 of 6,059), and 22.5% among Asian women (N=943 of 4,192). By comparison, the annual cervical cancer screening rate of the general population in California in 2010 was calculated to be 42.3%.

Table 1 presents the analysis of potential predictors of screening in the sample, controlling for age, race-ethnicity, and county type. In general, Asian (ARR=1.23), black (ARR=1.10), and Hispanic (ARR=1.11) women with severe mental illness were significantly more likely to have been screened for cervical cancer during the study period compared with their white counterparts (p<.001). Women ages 28–47 were significantly more likely than those ages 18–27 to have been screened (ARRs=1.31–1.32, p<.001). No significant difference in screening rates was found for women with severe mental illness who had a history of drug or alcohol use. When the analysis controlled for a history of drug or alcohol use, women with bipolar disorder, major depressive disorder, or generalized anxiety disorder were significantly more likely than those with schizophrenia to have been screened (ARRs=1.18–1.26, p<.001). Patients with evidence of health care use were significantly more likely than those without such evidence to have been screened (ARR=3.07, p<.001).

This study found that only 20% of women with severe mental illness received cervical cancer screening within the one-year study period, a screening rate less than half of that in the general population of women in California during the same period (42%) (12). The findings are concerning because these women were receiving public health services (in specialty mental health settings), but it appears that they did not receive preventive services as often as women in the general population. In addition, because women with severe mental illness may be at an increased risk of invasive cervical cancer because of the prevalence of risk factors, such as high rates of smoking and larger numbers of lifetime sexual partners (13–15), the lack of screening is particularly problematic.

The results indicate a need to prioritize women’s health screenings in specialty mental health settings, in addition to previously identified areas of focus, such as improving metabolic screening (8). One solution might include conceptualizing the specialty mental health clinic as the medical home and holding a “women’s mini-clinic” on site once a month. This might be akin to the primary care satellite clinics that were piloted over the past few years by the Primary and Behavioral Health Care Integration program of the Substance Abuse and Mental Health Services Administration. Another solution might involve creating a registry of cervical cancer screening status of all women in these specialty mental health clinics so that their providers could prioritize referrals to primary care or gynecology services.

Second, we found that cervical cancer screening rates differed among diagnostic groups of women with severe mental illness, such that women with psychotic disorders had significantly lower rates of screening than women with other major psychiatric diagnoses. The pronounced cognitive, psychosocial, and behavioral deficiencies often reported among persons with psychotic disorders may play a role in especially low rates of use of preventive health services.

We also found that women from minority racial-ethnic groups had significantly higher rates of cervical cancer screening than their white counterparts, with Asian women having the highest rate overall. This finding is interesting given that screening rates among Asian women have been previously reported to be lower than those among other racial-ethnic groups because of cultural and psychosocial beliefs, decreased access to care, and lack of knowledge about preventive care (9). Future research should aim to assess the validity of self-reported cancer screening data in this population to determine whether systematic underreporting is a factor contributing to this discrepancy.

Women with severe mental illness ages 28–47 were significantly more likely than women ages 18–27 to have had cervical cancer screening. In 2009, the American Congress of Obstetricians and Gynecologists published recommendations that cervical cancer screening begin at age 21. As a result of these guideline changes, women ages 18–21 might have lower rates of cervical cancer screening. However, this finding may also reflect transitioning guidelines during the time of data collection; in 2012, national organizations reached a consensus to increase the routine cervical cancer screening interval from two to three years (2).

Not surprisingly, use of primary care was an important mediating factor that facilitated higher rates of cervical cancer screening, because such tests are typically performed in a primary care environment. This finding is consistent with our prior studies of screening for other health indicators (11). As such, improving primary care access for this vulnerable population may be critical for improving overall screening rates. Although improving access may seem straightforward, more work is needed to address previously identified barriers to use of primary care for this population, including difficulties with transport and access, adverse experiences, fear of mistreatment, embarrassment, and history of sexual assault (5).

This study had several limitations. First, patients dually eligible for Medicaid and Medicare were excluded, and thus screening rates may not reflect rates for the entire population with severe mental illness. In addition, data on cervical cancer screening not billed to Medicaid was not captured, which could have led to an underestimation of screening rates. This study was limited to one year of retrospective cervical cancer screening data—instead of three years—and included only California women with severe mental illness in contact with the public specialty mental health system. These factors may affect generalizability. In addition, our comparison group was the general California population, rather than California Medicaid recipients without severe mental illness. The comparison group might have received better cervical cancer screening in general, compared with the Medicaid population; thus the findings may overstate disparities in care received by people with severe mental illness. In addition, the screening rates of the comparison population were self-reported, rather than derived from billing data, which also affected the accuracy of the comparison. Finally, we were unable with the data that were available to address the impact of poverty (including transportation costs) and how health care decisions might have been made (for example, prioritizing mental health care over preventive screening).

Cervical cancer screening remains suboptimal among women with severe mental illness served in the public health system. The alarmingly low rates of screening among women with severe mental illness represent an unmet public health need, and action to alleviate this disparity should be a priority. Future research should aim to better understand the inadequate nature of cancer care in this population, working not only to reduce the cervical cancer–related mortality gap that exists between persons with severe mental illness and the general U.S. population but also to remove access barriers to care.

This study was initiated during a state quality improvement project to integrate primary care and mental health care, called the California Mental Health Care Management Program (CalMEND). The project was a collaboration between the California Department of Mental Health and the Pharmacy Benefits Division of the California Department of Health Care Services. Study investigators acknowledge CalMEND staff for their assistance in combining administrative databases without compensation.