Health Systems

---

KPNW's Safety Net for Preventive Services: The Challenge of Reaching the Unscreened | to pdf >>
By Nancy H. Stevens, PhD

In 1991, a Middle Management Development Project (MMDP) proposed that Kaiser Permanente Northwest (KPNW) strengthen its prevention program by creating a Safety Net. The proposal called for the prevention Safety Net to ensure that members at greatest risk, eg, the unscreened, would receive services known to be effective in decreasing the risk of morbidity and mortality.

In 1994, a Prevention Steering Committee selected breast and cervical cancer screening as initial Safety Net interventions as these were known to be cost-effective and to have predictable screening intervals. A dual strategy was developed: unscreened members would receive an outreach letter; in addition, those who had clinical encounters would receive a verbal reminder from their clinician that they were overdue for screening.

Data suggest that the Safety Net has contributed to improved screening performance. Moreover, preliminary data from KPNW's Tumor Registry suggest that we are finding a greater incidence of invasive cervical cancers that may have gone undetected without the Safety Net initiative. KPNW believes the Safety Net has enhanced preventive services for the Region. The Safety Net should prove to be a potent strategy for other preventive and population-based screening services.

---

"By the year 2000, some experts claim it will be standard practice for health plans to record in a data base what happens during the course of a patient's visit and what the outcomes are. Providers and employers alike will measure quality in part by this data, which will show how many patients receive preventative health measures, such as immunizations and cancer screenings."

-- Spectrum; Summer/Fall, 1991

---

But also in 1991, a Middle Management Development Project (MMDP) Team proposed to KPNW senior management a system to identify and deliver screening and other preventive services to members at highest risk.¹ This proposal coincided with mounting expectations that in 1993 we would be required to report regional performance on certain Program outcomes such as breast and cervical cancer screening rates. The convergence of these events led to the Region's Prevention Steering Committee deciding to sponsor and thereby strengthen centralized prevention services through the Safety Net.

In the ensuing years, 1994-1997, KPNW's Safety Net has evolved as a centralized function to support clinician delivery of prevention services. The Safety Net has fulfilled much from its original objectives and has:

• Assured that effective prevention services are delivered to as many members as appropriate,
• Delivered effective prevention services cost-effectively by focusing outreach and education efforts on women not receiving prevention services on their own,
• Created a partnership between member and health plan to maintain women's health,
• Protected health plan from medicolegal risk by initiating and documenting outreach efforts to women not coming for prevention services on their own,
• Created a system to synthesize outreach needs for members by standardizing and improving messages delivered.

It remains for the Safety Net to extend to outreach relating to specific medical follow-up by gaining increased organizational support for other prevention and population-based services.

Development of the Safety Net Initiative
Applying a concept that would go far beyond a centralized outreach program, the KPNW Safety Net was initially proposed to ensure that members at risk would receive prevention services known to be effective in decreasing the risk of morbidity and mortality to within a certain desirable interval. Very few screening interventions met current Safety Net eligibility criteria--that is: a) were likely to be cost-effective; and b) had predictable screening intervals. In fact, only breast and cervical cancer screening, immunizations, and smoking cessation could meet the criteria. From among these screening interventions, breast and cervical cancer screening were selected as the first Safety Net interventions because of their known benefit, the ability of existing clinical programs to deliver services, and the existence of systems to support data analysis for both breast and cervical cancer screening.

Establishing Safety Net Parameters
Safety Net intervals are longer than is normally recommended for intervals between screenings. Still, these intervals generally fall within ranges acknowledged as safe for screening tests. For example, if KPNW were to adopt a typical Safety Net interval, "it might extend its recommended screening interval for cervical cancer from two to five years, as existing data indicate that after a negative result, the risk of developing invasive cancer within the next five years is highly unlikely."¹ With the longer interval, KPNW still has twoperhaps threeopportunities to screen women who do not seek screening during the known latency period between development of early dysplastic changes and onset of CIS (carcinoma in situ). In addition, the organization can realize substantial cost savings.

Existing guidelines for both breast and cervical cancer screening defined the parameters of desirable clinical service and maximum Safety Net intervals.

Guideline summary for breast cancer screening
Regular mammographic screening should be considered for all women 40 years of age and older. No definitive studies either prove or disprove that screening mammography in women ages 40-49 and over age 70 results in significant decrease in mortality from breast cancer. A woman should decide the frequency with which she has mammography screening on the basis of her individual risk for cancer by considering factors such as personal or family history.²

Safety Net Parameters--women ages 52-69 years who have not had a screening mammogram within the past two years.

"If a prevention service is clearly effective at preventing death and disability, a managed care system is both ethically and pragmatically required to deliver the service to as many of its members (of appropriate characteristics) as is justifiable within economic bounds. A common approach to this problem involves a campaign to maximize the receipt of service, and the shortening of the recommended cycle of delivery of the service in order to maximize the opportunities for delivering it. This approach is both highly inefficient (that is, expensive) and not very effective at reaching the underserved." -- Tom Vogt, MD,MPH Prevention Steering Committee, 1994

Guideline summary for cervical cancer screening
Pap smears should be repeated annually until the patient has three annual negative smears. Thereafter, recommended screening intervals extend to two to three years. All women between the ages of 20 and 70 years who have an intact uterus should be screened at least every three years.³

Safety Net Parameters--women ages 21-69 years who have not had a Pap smear within the past three years.

Initially, clinicians expressed concern about two aspects of the Safety Net planning. First, the concept of a maximum interval between screenings was unfamiliar; previous strategies to improve screening services had been to shorten, rather than to lengthen, the recommended interval cycle. More frequent screenings, it was assumed, could improve probability of early detection. But shortening the interval cycle is a much more costly approach to preventive care and all but ignores the screening needs of members who do not access the health care system.

Second, concern was expressed that the criteria for inclusion in the Safety Net did not include other behavioral, familial, or personal risk factors. But this concern overlooked a modest but crucial principle: the greatest risk factor for a condition that has an effective screening test is failure to be screened.

The Prevention Steering Committee addressed these concerns from an epidemiologic perspective and was able to maintain its original focus for the Safety Net.

Developing an Outreach Strategy
The aim of the outreach effort is to ensure that as many as are willing can receive services within the prescribed Safety Net intervals. For KPNW to reach members who were not inclined to seek clinical services in a given year (about 35% of members), an outreach component would be essential. Outreach was defined as a direct contact by KPNW to members--either by letter or phone call--to encourage those within the Safety Net to obtain screening for breast or cervical cancer. Interest in improving access and participation in cancer screening programs was growing at about this time,^4,5 and three KP research studies^6-8 were assessing the impact of outreach efforts to encourage women to be screened for breast and cervical cancer at recommended intervals. Their preliminary findings showed that about one fourth of subjects responded favorably to one or more contacts (21-37% of women overdue for mammograms; 20-34% of women overdue for Pap smears). Somkin et al recommended the "use of patient reminder letters as a first step in a mammography or Pap smear screening outreach program."^7,9

Developing an Inreach Strategy
The term "inreach" here means communication with the member at the time of a clinical visit to let her know that she is overdue for preventive screening. The thrust of inreach is to deliver the needed services, either immediately or shortly thereafter at another appointment scheduled at the patient's convenience. Somkin's studies^7,9 had shown that a combined outreach and inreach strategy was more effective than any single strategy. Recipients of a reminder letter and a chart note reminder were more likely to obtain the recommended screening in the following six months than women who received only the reminder letter. The weakness in the inreach strategy appeared to be that alerts and triggers to the clinician are only effective for women who seek appointments. With evidence from the three research studies showing that even a single direct contact with patients helped elicit desired behavior and that two contacts were more effective still, KPNW determined to develop a strategy that would employ both outreach and inreach components.

Exclusions
Vogt's⁸ study findings, although similar to those of the Somkin studies^7,9 in that both revealed the importance of outreach messages and multiple contacts, uncovered another important, if unexpected, finding. Direct contact with women in the Safety Net provided all-important exclusion informationthat is, reasons why a woman would not, or could not, be screened. KPNW determined to document exclusion information as a component of its screening strategy.

Infrastructure to Implement the Safety Net
KPNW had most of the necessary clinical and technical support for a Safety Net function. Medical Economics could provide analytic support; Information Services, access to data systems; Tumor Registry, a home for the Safety Net data; and primary care clinicians, the necessary clinical services. What the organization lacked, howeverat least at the onsetwas the personnel to mount a telephone outreach program. So, although telephone contact was preferred, KPNW opted for patient reminder letters. Letters are sent to women the first year they appear in the Safety Net but not thereafter, as Vogt et al⁸ had demonstrated only incremental improvement in screening behaviors from multiple letters.

Developing the capacity to identify women in the Safety Net at the time of a clinical encounter was problematic at first because KPNW lacked the necessary clinical information systems to support this need. But in October 1996, shortly after the Safety Net initiative was developed, a "prevention" screen was introduced into the KPNW Results Reporting System that could electronically summarize the screening history of members (Fig. 1).

By mid-year 1996, KPNW had introduced inreach in all primary care offices. When breast and cervical cancer screening history was absent from the Results Reporting System (indicating no internal record of an examination), clinical assistants were expected to ask prescribed questions related to the patient's history and to note exclusions.

Characteristics of Women in the Safety Net
Once the Safety Net had been in place for a few years, we began to notice some distinct characteristics of unscreened women and to detect some changes in screening performance within the Region. We began to appreciate the unique qualities of women who, for whatever reason, have remained unscreened in spite of national and local efforts to emphasize the importance of screening and early detection of cancer. Although the characteristics of unscreened women are now being regularly reported in the literature,^10-12 we have the opportunity to both substantiate research findings and offer new insights from an applied, managed care setting.

Women examined in the Safety Net are drawn from a pool of members who have met local guideline specifications for gender and age as well as criteria for continuous enrollment in the Kaiser Foundation Health Plan. After all eligible women are identified, those with documented permanent exclusions are eliminated from the pool, leaving a group of unscreened women who will be recipients of outreach and inreach efforts for the duration of that year.

In 1997, women in the Safety Net made up 23% of women eligible for breast cancer screening and 22% of women eligible for cervical screening services.

Women with No Clinical Encounters
One characteristic we found in most women in the Safety Net is their infrequent pattern of primary care visits. This pattern is particularly true for women in the cervical cancer Safety Net.

A significant portion of women in the Safety Net had no clinical visit during the previous year (52% for cervical cancer screening; 40% for breast cancer screening). This portion is far greater than that reported for the entire KPNW population (35%). These data support previous findings that women who do not get screened do not have regular health care visits.¹² The fact that nearly half of women at risk have not sought an appointment within a year suggests that some women perceive barriers to seeking primary care services, particularly women who remain unscreened for multiple years. This subpopulation, whom contact remindersno matter how numerous and no matter how conveyeddo not convince, poses a unique challenge, and we must consider different strategies to better understand and reach this population.

Women who Remain Unscreened after Clinical Encounters

	Breast Ca Screening	Cervical Ca Screening
1996	2,317 (28%)	5,325 (37%)
1997	2,202 (26%)	3,316 (23%)

One group we have closely watched are women with a primary care visit who do not get screened. These women either slip through our inreach efforts or defer the invitation to be screened. Our data systems do not allow us to determine the precise reasons screening does not occur during the course of a clinical encounter, but our exclusion data do offer us some explanation about why women don't get screened (Table 1). However, we are unable to determine whether the decrease in cervical cancer screening for women with clinical encounters is due to improved intervention efforts or to better documentation of permanent exclusions among these women.

Women Affiliated with a Primary Care Clinician
More than two thirds (68%) of all our KPNW members report an affiliation with a specific primary care clinician. Women in the cervical cancer Safety Net are less likely to affiliate (60%) than the member population, although the percentage varies greatly among medical officesfrom as low as 46% to as high as 84%. For reasons we do not understand, the percentage of affiliated women within the breast cancer Safety Net (70%) more closely matches the affiliation rate in the member population.

Women with Longevity in the Safety Net

	Breast Ca Screening	Cervical Ca Screening
First year in Safety Net	2,527 (40%)	4,513 (37%)
Second year	1,011 (16%)	2,561 (21%)
Third year	2,780 (44%)	5,123 (42%)
Total	6,318 (100%)	12,197 (100%)

We have found that this year, 1998, the largest proportion of women in the Safety Net have been there since the list was first generated in 1995.

The fact that we have a subset of women who are chronically unscreened in spite of our attempts to reach them is somewhat discouraging but an important factor to acknowledge as we continue to improve Safety Net efforts. A quick examination of these women by age and visit history shows little differentiation by number of years in the Safety Net. Insight into the "chronically unscreened" is not provided by the research literature. Given our ability to retrieve archived data on these women, we hope to segment unscreened women by the length of time they have remained unscreened and learn more in order to better understand them.

Impact of the Safety Net
The most important goal of this initiative was to improve delivery of breast and cervical cancer screening services. As outreach was implemented in early 1996 and inreach implemented later that year, we did not anticipate seeing any impact on performance until the end of 1997 at the earliest.

Screening Rates in the Safety Net

	Breast Ca Screening	Cervical Ca Screening
1996	2,793 (34%)	3,058 (21%)
1997	2,733 (32%)	2,754 (19%)

The screening rates for women in the Safety Net have been monitored for the two years of implementation. The figures below indicate that the screening rates of women in the Safety Net remained steady in 1996 and 1997. These screening figures are not unlike the screening ranges documented in the previously cited research studies.^10-12 They also support Somkin's finding that increased screening rates after interventions are generally lower for cervical cancer than for breast cancer when similar techniques are used.

The minor changes between 1996 and 1997 may be statistical variation or may reflect a true decrease in the screening rate. This population may be becoming harder to convince that screening is important as a greater proportion of women in the Safety Net have now been there for multiple years, and we have improved our ability to detect permanent exclusions and avoid misclassification of appropriateness for screening.

Regional Screening Performance
Even before the Safety Net was introduced, KPNW was collecting HEDIS (Health Plan Employer Data and Information Set) data for both breast and cervical cancer screening. In 1996, we began to measure performance according to KPNW's own local specifications, thus broadening the screening specifications for breast and cervical cancer to include all women to age 69 years (Table 2). (HEDIS screening data were originally limited to commercially enrolled women up to age 65 years.)

As the regional performance measures show, the Safety Net appears to contribute to improved screening performance in 1997-98. Some believe that improvements may derive from more systematic documentation of exclusions, particularly for Pap smear testing, rather than from an increase in screening the previously unscreened.

Early Detection of Cancer
The Safety Net must not only improve screening rates but must also detect cancer at a curable stage, particularly among women who may be at increased risk. The data compare the results of cancer screening in the Safety Net population to figures from the KPNW population as a whole, using incidence figures from the KPNW Tumor Registry for 1997.

Breast Cancer
In 1997, 24,826 (77.8%) KPNW women between the ages of 52 and 69 years were screened for breast cancer. Of these, 2733 were in the Safety Net. That same year, 184 analytic cases of breast cancer were diagnosed in these KPNW women, 33 of whom were among women in the Safety Net. When we adjust the rates of cancer per 100,000, we find that the breast cancer rate is similar to that in the KPNW member population (Table 3). The difference in cancer rates between the two populations is minor and may be due to the few cancers found in the Safety Net as well as to the screening efforts that have continued in the KPNW Program for many years.¹³

Cervical Cancer

In 1997, 53,620 women between the ages of 21 and 69 years were screened for cervical cancer; of these, 2754 were in the Safety Net. In the same year, 14 cases of cervical cancer were diagnosed in women in the Safety Net, half of them in situ and half invasive (three localized, three regional, and one distant). The age-adjusted rates show the rate of invasive cancers was more than ten times the rate in the KPNW member population, although the in situ cancers did not show a similar elevated rate (Table 4).

We cannot be certain why the rate of invasive cervical cancer is high in the previously unscreened women, particularly without a comparable increase in in situ disease. Our findings indicate that risk factors for the development of cervical cancer include earlier onset of sexual partners, more partners, other sexually transmitted diseases, cigarette smoking, and lower socioeconomic class. Some or
all of these factors may also keep women from attending screening.

The cancer rates should be regarded as preliminary because they are based on relatively small numbers and on only one year's data. They must be interpreted cautiously and be followed over time.

Discussion
Learnings associated with this project go beyond the quantitative data we've discovered about unscreened members and our ability to influence their screening behaviors. Like most other large-scale projects, the Safety Net has given us a number of organizational challenges and unanticipated learnings. These experiences have influenced ongoing design of the Safety Net.

Several of the Program goals set out in 1994 have been achieved. The Safety Net has delivered on its goal to improve effective prevention services, at appropriate intervals, to as many members as possible. The Safety Net also affords protection to KPNW from medicolegal risks by initiating and documenting outreach efforts to those at risk. One unanticipated and notable achievement has been our ability to monitor cancer outcomes among women in the Safety Net.

However, the Safety Net has been only partially successful thus far in delivering prevention services in as systematic and cost-effective a way as possible. And we have encountered several systemic issues that require our continued attention and negotiation with other KFHP and medical departments. Some of these include:

Exclusions
Exclusions for performance measurement are generally specified by HEDIS or another sponsoring organization. There is, however, no such internally agreed upon use of exclusions for the Safety Net, and ideas about the best way to treat Safety Net exclusions continue to generate considerable debate among clinicians.

KPNW has taken the position that women with permanent exclusionsbilateral mastectomy, hysterectomy, membership lapse, permanent medical limitations, or terminal illnessare ineligible for the Safety Neteven if those exclusions are not recognized in HEDIS performance measurement. In contrast, temporary exclusionsrefusal by women to undergo tests, evidence that tests were performed outside KPNW, existence of a temporary medical conditiondo not eliminate a woman from the Safety Net list.

Access for Women in the Safety Net
In spite of our efforts to facilitate appointment-making for women who respond to the outreach letter, we have been frustrated in our slow progress to assure their quick access to screening. We have only partially succeeded in finding a consistent, reliable method of identifying these women when they request clinical appointments. We must continue to work to eliminate any barriers to these women when they call in or arrive for appointments.

Overscreening
One discovery of the Safety Net was the extent to which we were screening women who had documented hysterectomy. Although regional guidelines state that women do not need Pap smear screening if they have had a hysterectomy for benign conditions, we found that in one of our local markets, more than half the women with documented hysterectomy also had had a Pap smear within the past three years. This finding raised questions about whether we are overexcluding or overscreening women. Women excluded from screening have expressed confusion about their need for regular gynecologic examinations. As a result, we are modifying our communication to women, distinguishing better between cervical cancer screening and other screening examinations.

Safety Net Timeline
1991	Middle Management Development Team proposes "Safety Net: A Centralized Risk Registry for Prevention and Early Detection Outreach" to senior managers
1994	Safety Net Initiative adopted by Prevention Steering Committee
1995	Infrastructure developed; First Safety Net list generated; Outreach pilot tested
1996	Outreach initiated for all women needing "Mamms" and "Paps"; Inreach initiated (mid-year)
1997	First full year implemented; First outcomes documented
1998	Safety Net implementation maintained

Uncoordinated Outreach
A continuing challenge has been KPNW's efforts to coordinate its outreach contacts. Currently, a patient who is in both the Safety Net and the Diabetes Registry, for instance, may receive separate reminder calls or letters instead of one. Not only is this an inefficient use of resources, but it also alienates members, who perceive that the right hand knows little or nothing about what the left hand is doing.

Complexity of Unscreened Population
Until recently, we have considered unscreened women as a single entity and the strategy for promoting breast and/or cervical cancer screening as the same. However, we may not be able to continue this assumption because fewer of these women (<10%) require both services, and women needing only cervical cancer screening are emerging as distinct in several ways. Besides the fact they are a much larger cross-section of our membership, they are more likely to be excluded from future screening, are less likely to be affiliated with a primary care physician, have fewer clinical encounters with the health care system, and seem less likely to respond to our attempts to encourage screening. The growing public awareness of breast cancer and media appeal for women to get mammograms is one possible explanation. Another may be that cervical cancer is not perceived to be as much of a threat. All these factors must be considered as we examine our continued participation in Safety Net activities.

Conclusion
Our preliminary findings give us a new appreciation for the complexity of influencing screening behaviors, particularly among women who are continually resistant to our outreach and inreach efforts through the Safety Net. Reasons for this resistance, probably due to a diversity of demographic, psychosocial, and organizational factors, will need to be better understood in the future if we are to identify new strategies to reach unscreened members. Our experience has influenced not only design and implementation of this single Safety Net initiative but the organizational systems that support KPNW's clinical service delivery. We are encouraged by the potential of the Safety Net as a potent strategy to better understand the processes and outcomes of preventive as well as other population-based screening services--and all before the year 2000!

Acknowledgments: The author acknowledges members of the MMDP Team #8 from KPNW, August 1991, who formulated the initial goals and planned early development of the Safety Net: Tom Vogt, MD, MPH; David Moiel, MD; Steve Gordon, MD; Mary Anne Hannibal; Jamie Forsythe; David Stokey.

Several people made major contributions to further development and implementation of the Safety Net initiative: Deborah Harris; Beverly A. Battaglia; Andy Glass, MD; Belle Slesh; Danielle Engels.

---

References
1. Middle Management Development Program (MMDP). Safety Net: A Centralized Risk Registry for Prevention and Early Detection Outreach, Kaiser Permanente, Northwest, Team #8. 1991.
2. KPNW Clinical Practice Guideline. Breast Cancer Screening. Portland, Oregon 1995.
3. KPNW Clinical Practice Guideline. Cervical Cancer Screening. Portland, Oregon 1996.
4. Vernon SW, Laville EA, Jackson G. Participation in breast screening programs: a review. Soc Sci Med 1990;30:1107-18.
5. Ornstein SM, Musham C, Reid A, Jenkins RG, Zemp LD, Garr DR. Barriers to adherence to preventive services reminder letters: The patient's perspective. J Fam Pract 1993;36(2):195-200.
6. Binstock M, Hackett J. Pap Smear Outreach: A Randomized Controlled Trial of Four Interventions. KP Medical Care Program, Southern California, 1993.
7. Somkin CP, Hiatt RA, Tillinghast S, Larson P. Computerized Reminders for Mammograms and Pap Smears, KP Medical Care Program, Interregional Health Education Conference, San Diego, California October 1995.
8. Vogt TM, Glass AG, LaChance PA, Love P. Safety Net Screening for breast and cervical cancer. Personal communication. Center for Health Research, KPNW. 1995.
9. Somkin CP, Hiatt RA. Hurley LB, Gruskin E. Ackerson L, Larson P. The effect of patient and provider reminders on mammography and Papanicolaou smear screening in a larger health maintenance organization. Arch Intern Med 1997;157:1658-64.
10. Potter SJ, Mauldin PD, Hill HA. Access to and participation in breast cancer screening: a review of recent literature. Clin Perform Qual Health Care 1996;4:74-85.
11. Weinberg AD, Cooper HP, Lane M, Kripalani S. Screening behaviors and long-term compliance with mammography guidelines in a breast cancer screening program. Am J Prev Med 1997;13:29-35.
12. Calle EE, Flanders WD, Thun MC, Martin LM. Demographic predictors of mammography and Pap smear screening in US women. Am J Public Health 1993;83:53-60.
13. Glass AG, LaChance P, Vogt TM. Eighteen years of mammography experience in an HMO: 1975-1992. HMO Pract 1996;10(4):184-9.

CMI Endorsed Successful Practice: Cervical Cancer Screening Program The Care Management Institute (CMI) has recently endorsed the Northwest Region's Cervical Cancer Screening Program, "Safety Net," as a successful practice. CMI established the Successful Practices Program as a means to identify highly effective and innovative programs within Kaiser and to help facilitate the program's transfer and adoption to other sites. CMI evaluates the quality of the program, ensures the approach is supported by evidence, assesses potential problems, highlights key program requirements and success factors, and summarizes the program in a succinct, actionable format. Endorsement of a successful practice refers to the process CMI uses to determine the program's merit. Endorsed successful practices are subsequently disseminated via CMI's implementation network to encourage rapid adoption and sustained implementation throughout Kaiser Permanente.