Epidemiology/Health Services/Psychosocial Research O R I G I N A L
A R T I C L E
Risk and Protective Factors Associated With Screening for Complications of Diabetes in a Health Maintenance Organization Setting LISA PAYNE SIMON, MPH ANN ALBRIGHT, PHD, RD MICHAEL J. BELMAN, MD, MPH
ERIC TOM, MBA, MSC JEFFREY A. RIDEOUT, MD, MA
OBJECTIVE — To identify risk and protective factors associated with screening for complications of diabetes, we performed a cross-sectional study of 3,612 diabetic members enrolled in CaliforniaCare, a large network-model health maintenance organization (HMO). RESEARCH DESIGN AND METHODS — We used the Health Plan and Employer Data Information Set (HEDIS) 3.0 technical definition to identify all members (aged 31 years) receiving any diabetes medication(s) during a 12-month period. Using a telephone survey instrument, identified members were interviewed about their diabetes care and screening, patient, and provider history. Survey data were supplemented with HMO claims and demographic information. Multivariate analysis was performed to identify demographic, clinical, and utilization characteristics that affect the odds of diabetic members receiving annual retinal examination, foot examination, and HbA1c testing. RESULTS — While results varied by screening category, the odds of obtaining screening were higher for diabetic members who were older, spoke English, received diabetes nutrition counseling, visited a diabetes specialist physician, belonged to a diabetes association or support group, used insulin, performed glycemic level self-examination at least once a day, and had higher overall prescription drug use (suggesting higher comorbidity). Since this study is a cross-sectional review, these results do not imply a cause-and-effect relationship between dependent and independent variables. CONCLUSIONS — Results of this study suggest barriers, risks, and protective factors associated with screening for complications of diabetes. Diabetic members who do not possess these characteristics may be at increased risk. Diabetes Care 22:208–212, 1999
iabetes is a chronic illness that, if not properly managed, can lead to significant morbidity and mortality. According to Centers for Disease Control and Prevention (CDC) estimates, 15.7 million adults (5.9% of the U.S. population) have
D
diabetes, and among these, 10.3 million have been diagnosed (1). Diabetes prevalence increases with age. Among people aged 65 years, 18.4% have diabetes, compared with 8.2% of people aged 20 years and 1% of people aged 20 years (1). In
From Blue Cross of California (L.P.S., M.J.B., E.T.), Woodland Hills; University of California at San Francisco–Stanford Health Care (L.P.S., J.A.R.), San Francisco; and the California Diabetes Control Program (A.A.), Sacramento, California. Address correspondence and reprint requests to Michael J. Belman, MD, MPH, Regional Medical Director, Medical Quality Management, Blue Cross of California, 21555 Oxnard St., Suite 6J, Woodland Hills, CA 91367. E-mail:
[email protected]. Received for publication 2 June 1998 and accepted in revised form 7 October 1998. Abbreviations: CCHRI, California Cooperative Healthcare Reporting Initiative; HEDIS, Health Plan and Employer Data Information Set; HMO, health maintenance organization; ICD-9, International Classification of Diseases, Ninth Revision ; NCQA, National Committee for Quality Assurance; OR, odds ratio. A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.
208
1995, diabetes was the seventh leading cause of death in the U.S. (1). Diabetesrelated medical costs (treatment and hospitalization expenditures) were $45 billion in 1992, and estimated total costs (including disability, work loss, and premature death) were $92 billion (1). Nearly 25% of direct medical costs—over $10 billion annually— are accrued during admissions stemming from long-term complications (2). Regular screening for diabetic retinopathy, nephropathy, glycemic control, and neurologic and vascular disease associated with diabetes is necessary for the prevention and control of disease complications (3). Diabetes care guidelines, including those from the American Diabetes Association and from the California Diabetes Control Program and Diabetes Coalition of California, recommend periodic eye examination; foot examination; HbA1c testing; microalbuminuria screening; urine examination; annual blood test for cholesterol, triglycerides, and HDL; routine glucose level self-examination; and other recommended screenings for the prevention of diabetes complications (4,5). Despite consensus surrounding these guidelines, a number of studies have revealed low rates of screening (6–10). Among health maintenance organization (HMO) populations, lack of compliance can also be inferred based on the National Committee for Quality Assurance (NCQA) Health Plan and Employer Data Information Set (HEDIS) results for diabetic retinal examination and HbA1c testing. National screening rate averages published by the NCQA revealed a 1997 diabetic retinal examination rate of 38% (10). Similarly, the California Cooperative Healthcare Reporting Initiative (CCHRI) revealed that the average health plan eye examination rate was only 37% in 1997 (up from 29% the year before) (10). CCHRI also found that only 22% of adult diabetic members ( 65 years of age) enrolled in California commercial health plans received at least one HbA1c test during 1996 (11). That same year, 10 California provider groups measured HbA 1c
DIABETES CARE, VOLUME 22, NUMBER 2, FEBRUARY 1999
Simon and Associates
testing rates within their populations, and a mean rate of 40% was observed (11). These findings suggest low levels of compliance, both nationally and in California, with guidelines recommending annual screening. The research objectives of the California Care Diabetes Outreach Program were to identify risk and protective factors associated with screening for complications of diabetes and to shed light on potential barriers to screening using a cross-sectional study design. Based primarily on experience gained through HEDIS data collection for diabetic retinal and HbA1c examinations, we hypothesized that specific member characteristics (including English as primary language, having seen a diabetes specialist physician, and prior diabetes education) might increase the odds of screening. In this study, analysis is limited to three categories of screening—eye examination, foot examination, and HbA1c testing—for which member screening status information was collected using a survey instrument. In addition to identifying risk and protective factors associated with variation in rates of screening for complications of diabetes, the Diabetes Outreach Program also had specific programmatic objectives. These included the following: make telephone contact with all diabetic members, assess members’ current screening status, educate members regarding screening and other diabetes care requirements, provide consultation to diabetic members regarding benefit coverage and available resources, facilitate provider contact for members with specific screening or care requirements, and identify and address members’ perceived barriers to screening. RESEARCH DESIGN AND METHODS Study population Consistent with the HEDIS 3.0 technical specification for the identification of adult health plan members with diabetes, this study included all diabetic members (aged 31 years) who were enrolled in CaliforniaCare as of 1997 and who had a pharmacy claims history indicating that they had filled at least one outpatient prescription for insulin or oral hypoglycemic medication during a recent 12-month period (12). This definition does not allow for the identification of diabetic members who are diet controlled. National estimates indicate
that 25–30% of the diabetic population is diet controlled. It is likely that a similar proportion of diet-controlled diabetic plan members are excluded from this study. Among diabetic members who take medication for their condition, comprehensive identification through pharmacy claims is possible, since 96% of members have combined medical and pharmacy coverage.
Statistical analysis Results frequencies were calculated for all survey, claims, and membership data collected for the study. Differences between Outreach Program respondents and nonrespondents were compared on all available demographic, clinical, and healthcare utilization variables. For continuous variables (age), nonpaired Student’s t tests were performed. For categorical variables (including members’ sex, primary language, region of residence, hospitalization for diabetes, and medication use), 2 or Fisher’s exact tests were performed. All analyses were performed using STATA statistical software (Stata, College Station, TX). Annual retinal examination, foot examination, and HbA1c testing were each evaluated using multiple logistic regression. Models included the presence or absence of screening as the dependent variable. Independent predictor variables included the following: member age, sex, primary language spoken, whether a diabetes specialist physician was seen (yes/no), whether a dietitian or nutritionist was seen (yes/no), diabetes education received (yes/no), insulin injection frequency (not at all, up to once a day, or more than once a day), type of diabetes medication used (oral, insulin, or both), number of 1997 inpatient claims with an ICD-9 code for diabetes (zero vs. one or more), and number of prescriptions filled (including all prescription drug categories as a proxy for comorbidity status). The prescription drug utilization distribution was divided into thirds: 34.3% of the study population (low utilizers) filled up to 14 prescriptions during a 12-month period, 32.1% (moderate utilizers) filled between 15 and 23 prescriptions, and 33.6% (high utilizers) filled 24 or more prescriptions in a 12-month period. For each regression model, we present adjusted odds ratios (ORs) and 95% CIs.
Data collection The Diabetes Outreach Program began on 1 October 1997 and was completed on 19 December 1997. The outreach consisted of a structured telephone interview. A standardized data collection instrument was used to record member responses to questions on diabetes-related health service utilization, care, and screenings (including eye examination, foot examination, HbA1c testing, and glycemic level self-examination). Surveys were conducted by trained interviewers. A team of clinicians was also available during phoning to provide consultation specific to members’ identified needs and requests for information and to facilitate provider referrals. Each of the 7,288 identified members was telephoned up to five times until contacted. Telephone contact information was unavailable for 2,110 members (29%). These members were contacted by mail and invited to participate in the study. A total of 3,628 members completed the survey during the 3-month study period. This statistic represents 50% of all eligible medication-controlled diabetic members and 70% of members for whom telephone contact information was available. Members’ screening status and information regarding diabetes treatment, outpatient utilization, and patient history were obtained via the Diabetes Outreach Program survey. Additional data were collected for all identified members using the Blue Cross of California membership, medical claims, and pharmacy claims databases. RESULTS — The following results were Membership database information included observed among the 3,612 study particithe following: member name, address, tele- pants: 77.9% reported a diabetic retinal phone number, age, sex, primary physi- examination within the past year, 65.2% cian, and medical group affiliation. Medical reported a foot examination, 89.0% reported claims database information included mem- HbA1c testing, and 78.2% reported monibers’ inpatient claims with a principal, sec- toring their glycemic level at least periodiondary, or tertiary diagnosis of diabetes cally. In addition, 16.4% of the sample (International Classification of Disease, Ninth reported having seen a diabetes specialist Revision[ICD-9] code 250.xx). Pharmacy physician, 63.6% had seen a dietitian or claims database information included mem- nutritionist, 41.5% had seen a diabetes edubers’ claims for diabetes medication and cator, 36% had attended diabetes classes, total prescriptions filled during a recent 12- and 10.5% belonged to a diabetes associamonth period (ending July 1997). tion or support group. Also, 76.5% agreed
DIABETES CARE, VOLUME 22, NUMBER 2, FEBRUARY 1999
209
Screening for complications of diabetes
Table 1—Characteristics that affect the odds of receiving annual retinal examination in 3,612 ature as factors associated with diabetes diabetic HMO members aged 31 years care and screening (1,6,13). Characteristic
OR (95% CI)
Age (per 5 years) Female sex Primary language is English Diabetes specialist (doctor) seen Dietitian/nutritionist seen Diabetes educator seen Attended one or more diabetes education classes Belongs to a diabetes association or support group Medication type (referent is oral) Insulin Both oral and insulin Frequency of glucose level examination (referent is “Not at all”) 1 per day 1 per day 1997 prescription drug use (comorbidity proxy; referent is low utilization) Moderate utilization High utilization
that educational outreach of this nature was what they expected from their health plan. To assess the generalizability of study findings, differences between Diabetes Outreach Program participants and members who could not be reached for interview were compared. It is noteworthy that among those diabetic members who did not participate, only 4.6% (n = 168) actually declined participation, refusing to complete the telephone survey. The remaining 95% could not be reached for interview. Results suggest that, compared with diabetic members who did not complete the survey, members who did complete the survey were older (mean age 51.8 vs. 50.2 years, P 0.001), were more likely to reside in northern California (north of and including San Jose and the San Francisco Bay area, 23.4 vs. 19.5%, P 0.01), had higher mean prescription drug use (all medications, 21.4 vs. 17.8 prescriptions per year, P 0.001), and had a higher rate of recent hospitalization for diabetes (13.2 vs. 9.9%, P 0.001). Observed differences suggest that, compared with members who did not participate, study participants may have had more serious diabetes complications and a greater number of comorbid conditions. While these results may limit the generalizability of study findings to all diabetic populations, they also suggest that risk and protective factors identified in this study may be of 210
P value
1.11 (1.05–1.16) 1.10 (0.92–1.31) 1.32 (1.02–1.70) 1.07 (0.82–1.39) 1.48 (1.22–1.80) 1.07 (0.83–1.37) 1.41 (1.10–1.81) 1.94 (1.33–2.83)
0.001 0.28 0.03 0.61 0.001 0.62 0.01 0.001
1.21 (0.94–1.55) 1.15 (0.76–1.74)
0.14 0.51
1.32 (1.07–1.62) 1.72 (1.31–2.26)
0.01 0.001
1.30 (1.06–1.60) 1.30 (1.04–1.62)
0.01 0.02
particular relevance to higher-risk diabetic populations. Multivariate associations, expressed as ORs with 95% CI, are summarized in Tables 1–3. Independent variables were selected based on the results of bivariate analyses and for their relevance in the liter-
Table 1 suggests that the following characteristics increase the odds of diabetic members receiving an annual retinal examination: increasing age (OR = 1.11, P 0.001), English as primary language (OR = 1.32, P = 0.03), having seen a dietitian/nutritionist for counseling (OR = 1.48, P 0.001), having attended diabetes education classes (OR = 1.41, P 0.01), belonging to a diabetes association or support group (OR = 1.94, P = 0.001), performing glycemic level self-examination at least once a day (OR = 1.32, P 0.01) or more than once a day (OR = 1.72, P 0.001), and both moderate (OR = 1.30, P = 0.01) and high (OR = 1.30, P = 0.02) comorbidity status (defined as moderate and high overall prescription drug use). Results presented in Table 2 suggest that the following characteristics increase the odds of diabetic members receiving an annual foot examination: increasing age (OR = 1.10, P 0.001), male sex (OR = 1.38, P 0.001), English as primary language (OR = 1.53, P 0.001), having visited a diabetes specialist physician (OR = 1.60, P 0.001), having seen a dietitian/nutritionist (OR = 1.63, P 0.001), insulin medication use (OR = 1.31, P = 0.01) or combined oral and insulin medication use (OR = 1.45, P = 0.04), performing glycemic level self-examination at
Table 2—Characteristics that affect the odds of receiving annual foot examination in 3,612 diabetic HMO members aged 31 years Characteristic
OR (95% CI)
Age (per 5 years) Male sex Primary language is English Diabetes specialist (doctor) seen Dietitian/nutritionist seen Diabetes educator seen Attended one or more diabetes education classes Belongs to a diabetes association or support group Medication type (referent is oral) Insulin Both oral and insulin Frequency of glucose level examination (referent is “Not at all”) 1 per day 1 per day One or more hospitalizations for diabetes during 1997 1997 prescription drug use (comorbidity proxy; referent is low utilization) Moderate utilization High utilization
P value
1.10 (1.06–1.15) 1.38 (1.19–1.61) 1.53 (1.21–1.93) 1.60 (1.27–2.01) 1.63 (1.38–1.94) 1.18 (0.95–1.46) 1.08 (0.87–1.33) 1.14 (0.87–1.49)
0.001 0.001 0.001 0.001 0.001 0.13 0.68 0.34
1.31 (1.07–1.62) 1.45 (1.02–2.06)
0.01 0.04
1.32 (1.09–1.59) 1.60 (1.26–2.02) 1.01 (0.81–1.27)
0.01 0.001 0.93
1.17 (0.98–1.40) 1.25 (1.03–1.52)
0.09 0.02
DIABETES CARE, VOLUME 22, NUMBER 2, FEBRUARY 1999
Simon and Associates
Table 3—Characteristics that affect the odds of receiving annual HbA 1c examination in 3,612 diabetic HMO members aged 31 years Characteristic
OR (95% CI)
Primary language is English Diabetes specialist (doctor) seen Dietitian/nutritionist seen Diabetes educator seen Attended one or more diabetes education classes Belongs to a diabetes association or support group Medication type (referent is oral) Insulin Both oral and insulin Frequency of glucose level examination (referent is “Not at all”) 1 per day 1 per day 1997 prescription drug use (comorbidity proxy; referent is low utilization) Moderate utilization High utilization
least once a day (OR = 1.32, P 0.01) or more than once a day (OR = 1.60, P 0.001), and high comorbidity status (OR = 1.25, P = 0.02). Moderate comorbidity status had a marginal effect on the odds of screening (OR = 1.17, P = 0.09). Results presented in Table 3 suggest that the following characteristics increase the odds that diabetic members have received an HbA1c test within the past 12 months: having seen a dietitian/nutritionist (OR = 1.35, P = 0.02), belonging to a diabetes association or support group (OR = 1.77, P = 0.02), and high comorbidity status (OR = 1.42, P = 0.02). Speaking English as a primary language reduced the odds of HbA1c screening (OR = 0.44, P 0.001). Receipt of diabetes education from a diabetes educator had a marginal impact on the odds of screening; however, this result did not reach statistical significance (OR = 1.32, P = 0.10). CONCLUSIONS — Results of this study suggest that among a large sample of HMO members with diabetes, the odds of receiving recommended annual screenings are, in general, higher for members who are older and English speaking, have received diabetes nutrition counseling, have visited a diabetes specialist physician, belong to a diabetes association or support group, perform glycemic level self-examination at least once a day, use insulin or both insulin and oral agents, and have higher overall prescription drug use (suggesting higher comorbidity). While these results do not
P value
0.44 (0.28–0.69) 1.22 (0.86–1.74) 1.35 (1.04–1.74) 1.32 (0.95–1.85) 1.16 (0.83–1.62) 1.77 (1.11–2.83)
0.001 0.26 0.02 0.10 0.38 0.02
0.92 (0.67–1.25) 1.17 (0.67–2.03)
0.58 0.58
0.77 (0.57–1.04) 0.83 (0.57–1.20)
0.09 0.31
1.21 (0.92–1.58) 1.42 (1.07–1.88)
0.18 0.02
imply cause and effect, these findings suggest that diabetic members who do not possess these characteristics may be at increased risk for complications due to inadequate secondary screening. Results suggest that for HbA1c examination, speaking Spanish as a primary language improved the odds of screening, whereas for retinal examination and foot examination, speaking English improved the odds of screening. According to Centers for Disease Control and Prevention estimates, Hispanics/Latin Americans are almost twice as likely to have diabetes as non-Hispanic whites of similar age (1). While it is difficult to explain the observed differences between measures, it may be that the higher rate of HbA1c screening observed is due to the provider community’s awareness of Hispanic populations’ higher risk. Understanding barriers, risks, and protective factors associated with screening for complications of diabetes may be useful in the development of interventions for improving diabetes-related quality of care. Barriers identified during the course of the Outreach Program have led to more explicit benefit language regarding retinal examination coverage by the plan and increased availability of diabetes education materials in Spanish. Also in 1998, Blue Cross of California is reinforcing its educational outreach activities toward specific at-risk populations identified through this research. Potential limitations to this study include reliance on member self-reporting of dia-
DIABETES CARE, VOLUME 22, NUMBER 2, FEBRUARY 1999
betic screening information. However, validation research has shown a relatively high concordance between medical chart–based evidence and member recall of ambulatory encounter events (14). A comparison study (in progress) of patient perceptions of screening for complications of diabetes versus screening events recorded in the medical chart revealed agreement between these sources for retinal examination and blood pressure examination, a higher reporting of HbA1c examination, and a lower reporting of foot examination in the medical chart (A. Albright, California Diabetes Control Program, personal communication). This evidence lends support to the validity and use of self-reported patient data in evaluating the occurrence of diabetes-related screening events. Because of the critical role of self-care in diabetes management, patient perceptions of care and screening received may also influence whether screening is sought or obtained. In preventing complications of diabetes, risk and protective factors associated with perceptions of screening may be as relevant as associations with validated screening events. A second limitation may be that, in accordance with the HEDIS 3.0 specification, diet-controlled diabetic members were not identified for inclusion in the study. As such, results may not be generalizable to all diabetic patients, but rather only to those patients who take oral or insulin medication for their diabetes and who are likely to have more serious diabetes and complications. Among this population, we also observed higher rates of screening among members whose medication use and glycemic level self-examination practices would suggest that they had more serious diabetes compared with the rest of the study sample. These factors may explain the higher than national average rates of screening observed for the sample, since it is likely that diet-controlled diabetic patients might also have fewer disease complications and therefore be less likely to obtain annual screenings. In summary, results of this study are generalizable to other at-risk diabetic populations who are likely to require more aggressive monitoring for adequate secondary prevention. Observed differences between Diabetes Outreach Program participants and members not reached for participation may also affect the representativeness of reported screening rates and other findings. For example, higher health service utilization rates observed among study participants 211
Screening for complications of diabetes
may increase the likelihood of this group obtaining the recommended screenings, so that reported rates may be somewhat higher for this group than for the entire CaliforniaCare diabetic population. On the other hand, 1.6% of the members contacted (n = 59) indicated that they did not have diabetes and were subsequently dropped from the study sample. Among nonparticipants, it was not possible to exclude individuals who did not have confirmed diabetes. Inclusion of misidentified diabetic patients in the nonparticipant sample may contribute to the lower rate of diabetes-related utilization observed among this group. While sample differences may limit the generalizability of study findings to all diabetic populations, we recognize that routine screening for complications of diabetes is recommended for all diabetic patients and that the results of this study suggest risk and protective factors associated with diabetic members who are potentially at greatest risk. A final potential limitation may be lack of inclusion of other key variables not available for analysis. For example, member income, ethnicity, and highest education level achieved, as well as information related to medical group and provider practice variation (e.g., whether or not the medical group provides case management through the use of a diabetes outreach nurse), might also shed light on barriers to screening. Through a cross-sectional design, this study identifies risk and protective factors associated with screening for complications of diabetes among 3,612 diabetic HMO members. Despite relatively uniform access to medical care through HMO membership, this study identifies large differences
212
in who seeks and obtains screening for diabetes complications. These differences may apply to other similar HMO member populations. Results of this study can be used by health plans and providers to identify potential barriers to screening, to evaluate outreach activity, and to design future interventions for the improvement of secondary prevention through screening for complications of diabetes.
Med116:683–685, 1992 5. Diabetes Coalition of California and California Diabetes Control Program: Basic Guidelines for Diabetes Care . Sacramento, CA, Diabetes Coalition of California, 1997 6. Jacques CHM, Jones RL, Houts P, Bauer LC, Dwyer KM, Lynch JC, Casale TSM: Reported practice behaviors for medical care of patients with diabetes mellitus by primary-care physicians in Pennsylvania. Diabetes Care14:712–717, 1991 7. Legorreta AP, Hasan MM, Peters AL, Pelletier KR, Leung KM: An intervention for enhancing compliance with screening recommenAcknowledgments — The authors are gratedations for diabetic retinopathy: a bicoastal ful to Kourosh Bahar for assistance in preparing experience. Diabetes Care20:520–523, 1997 data for analyses, Abraham Aronow, MD, MPH, 8. Sprafka JM, Frische TL, Baker R, Kurth D, and Therese Williams, MBA, for manuscript Whipple D: Prevalence of undiagnosed eye review, and Mary Spitzer, RN, Don Wentzel, disease in high-risk diabetic individuals. MD, Nancy Walker, RN, and Vicki Lelis for A rch Intern Med150:857–861, 1990 clinical and operations management of the Blue 9. Brechner RJ, Cowie CC, Howie LJ, Herman Cross of California Diabetes Outreach Program. WH, Will JC, Harris MI: Ophthalmic examination among adults with diagnosed diabetes mellitus. JAMA1270:1714–1718, References 1993 1. Centers for Disease Control and Prevention: 10. California Cooperative Healthcare ReportNational Diabetes Fact Sheet: National Esti ing Initiative: Report to CCHRI Participants: mates and General Information on Diabetes in 1996 Performance Results. Rohnert Park, the United States. Atlanta, GA, U.S. Dept of CA, O’Dell Publishing, 1997 Health and Human Services, Centers for 11. California Cooperative Healthcare ReportDisease Control and Prevention, 1997 ing Initiative: Report to CCHRI Participants: 2. Anderson D: Managed care meets the dia1995 Performance Results. 1996 betes management challenge. Business 12. National Committee for Quality Assurance: Health14 (Suppl. A):19–21, 1996 Health Plan and Employer Data Infor mation 3. Diabetes Control and Complications Trial Set, Version 3.0. Washington, DC, National Research Group: The effect of intensive Committee for Quality Assurance, 1998 treatment of diabetes on the development 13. Brancati FL, Whelton PK, Kuller LH, Klag and progression of long term complicaMJ: Diabetes mellitus, race, and socioecotions in insulin dependent diabetes mellinomic status: a population based study. tus. N Engl J Med329:977–986, 1993 Ann Epidemiol6:67–73, 1996 4. American College of Physicians, American 14. Brown JB, Adams ME: Patients as reliable Diabetes Association, and American Acadreporters of medical care process: recall of emy of Ophthalmology: Screening guideambulatory encounter events. Medical Care lines for diabetic retinopathy. Ann Intern 30:400–411, 1992
DIABETES CARE, VOLUME 22, NUMBER 2, FEBRUARY 1999