Purpose: This study evaluated the impact of an outpatient echocardiography utilization management (UM) program relative to a matched control (non-UM) group by measuring changes in utilization rates overall and within specific cardiac-risk subgroups.
Methodology: Administrative claims data for enrollees from five states were queried from the HealthCore Integrated Research Database (HIRD). Inclusion required continuous eligibility from Oct. 1, 2008, through Sept. 30, 2009, for pre-UM implementation and from Oct. 1, 2010, through Sept. 30, 2011, for post-UM implementation. Members were followed for one year before and one year after the implementation period; propensity-score matched; categorized as low, medium or high-risk for a cardiac event; and stratified into UM and non-UM cohorts. Changes in utilization rates were compared among the matched cohorts using the difference-in-difference approach; generalized estimating equations were used to adjust for relevant factors including age, gender and cardiac risk score.
Results: Members (N=2.5 million) were propensity-score matched (1:1) into UM and non-UM cohorts. Post- vs. pre-UM implementation, unadjusted utilization rates (echocardiography tests per 1,000 members) decreased for both the UM (16.9 percent) and non-UM (1.51 percent) cohorts, a difference of –15.42 points; the low risk subgroup had the greatest difference between UM and non-UM trends (–20.14 points) followed by the medium (–17.32 points) and high risk (–12.5 points) subgroups. Post implementation, 4.46 tests were avoided per 1,000 members in the UM cohort. After adjusting for relevant factors, the reduction in the UM group was estimated at –15.7% or 3.05 avoided tests per 1,000 members. In the low risk subgroup, 2,102 (49 percent of total) echocardiographs were avoided, followed by 1,302 (30% ) in the medium and 917 (21%) in the high risk subgroups.
Conclusion: UM for outpatient discretionary echocardiography appeared to reduce testing rates significantly after controlling for factors including age, gender, and risk score. Utilization avoidance was greatest in the low and medium risk groups. These findings add new insights in a field where there is a lack of knowledge regarding appropriate utilization by risk category.
Increasing utilization and high geographic variability drive concerns about the clinically inappropriate use of echocardiography and have resulted in the implementation of utilization management (UM) programs for many commercial health plans (Duszak 2012, Levin 2005). UM incorporates clinical review methods and tools to assess the appropriateness of health care services for specific patients and is designed to produce the best outcomes and optimized resource utilization (Duszak 2012, Hendel 2008, Otero 2006, Wickizer 2002). UM is used to guard against overutilization (Otero 2006, Handel 2008). Also, UM may avoid additional unnecessary follow-up testing when results are inconclusive. UM programs often are required for precertification and prior authorization (Weiner 2005, Wickizer 1995, Wickizer 2002).
Common UM models for medical imaging services include radiology benefits managers (RMBs) and real-time decision support (DS) tools. (Duszak 2012) These UM models can be mandatory, enforced through reimbursement consequences, or voluntary. Prospective clinical review employing clinical guidelines based on appropriate-use criteria developed by the American College of Cardiology Foundation and other relevant literature are used increasingly to manage echocardiography (Bhatia 2012, Mansour 2012, Parikh 2012).
Health care expenditures have been growing rapidly toward 20 percent of the United States gross domestic product (CBO 2012). Spending on diagnostic imaging outpaced all other medical services over the last two decades (Iglehart 2009). The Medicare Payment Advisory Commission (MedPAC) reported that expenditures doubled for imaging services, from $6.6 billion to $13.7 billion from 2000 to 2005, and the utilization rate of advanced diagnostic imaging by Medicare outpatients increased 72.7% (Levin 2011). Estimates of the total spent on imaging among commercial insurers were not readily available.
Recently, the growth rate for imaging services has slowed substantially (Duszak 2012, Levin 2012, Levin 2010, Levin 2011). The compound annual growth rate (CAGR) for non-invasive diagnostic imaging grew by just 1.4% from 2005 to 2008 (Levin 2011). Medicare data indicate that the volume of imaging services decreased by 2.5 percent from 2009 to 2010 (Med Pac 2012).
The current peer-reviewed literature demonstrating the effectiveness of UM programs for outpatient imaging is limited, especially on whether avoided utilization occurs among particular risk groups or across all categories of patients.
This large-scale population analysis of a mandatory UM program attempted to address this knowledge gap. The design included one-to-one propensity score matching based on claims-derived cardiac risk status, which increased the comparability of the case and control cohorts. Members were categorized according to their cardiac risk levels, and clinical elements were used to validate claims-based cardiac risk assessment.
The objective of this study was to evaluate changes in echocardiography utilization rates between managed (UM) and unmanaged (non-UM) members during the post-implementation (follow-up) period compared with the pre-implementation (baseline) period for the entire study population and for the low, medium and high cardiac risk subgroups.
Data sources and study design
This echocardiography clinical review program was implemented with an educational focus (without claims compliance enforcement) by AIM Specialty Health (AIM) in 2008 with the goal of managing discretionary diagnostic cardiac imaging. By the fourth quarter of 2010, the program was used for claims compliance and incorporated pre-authorization requirements for outpatient echocardiography services. The clinical review criteria developed for the program were based on American College of Cardiology Foundation’s appropriate-use criteria and other relevant literature (AIM 2012, Bonow 2012, Douglas 2011, Fleisher 2007, Mieres 2005, Warnes 2008).
Study population and design
Administrative medical claims data were queried from the HealthCore Integrated Research Database (HIRD), a repository of fully adjudicated medical, pharmacy and laboratory claims data for approximately 43 million Blue Cross and Blue Shield health plan members across 14 states in all geographic regions.
This observational cohort study included claims for outpatient cardiovascular imaging from Oct. 1, 2008, through Sept. 30, 2011, in Indiana, Kentucky, Missouri, Ohio, and Georgia for both UM and non-UM members. Echocardiography procedures were identified using Current Procedural Terminology, Fourth Edition (CPT–4) codes for echocardiography: 93303, 93304, 93306, 93307, 93308, 93312, 93315, 93318, 93350 and 93351.
Data for the cohorts were compared for the 12-month pre-UM implementation or baseline Oct. 1, 2008, through Sept. 30, 2009, and the 12-month post-UM-implementation or follow-up from Oct. 1, 2010, through Sept. 30, 2011. The implementation time between the baseline and follow-up periods was considered a silent period and excluded from the analysis for consistency.
Researchers had access to only a limited data set; strict measures were observed to preserve anonymity and confidentiality and to ensure full compliance with the Health Insurance Portability and Accountability Act (HIPAA) of 1996.
Health plan members from the five target states were included in the UM group if they were fully insured and were continuously eligible during the baseline period (Oct. 1, 2008, through Sept. 30, 2009) and during the follow-up period (Oct. 1, 2010, through Sept. 30, 2011). In addition, members were required to belong to specific employer groups that were subject to the AIM echocardiography UM program during the post-implementation period.
The non-UM cohort was required to meet all the eligibility requirements except for the fully insured requirement during the study time frame and did not belong to any of the employer groups that were subject to AIM echocardiography UM program in the post-implementation period.
Members with missing age and gender information were excluded, as were those with Medicare Supplemental, Medicare Advantage, or Medicaid coverage.
Propensity score matching
To minimize group bias and approximate the robustness of a randomized trial in an observational study setting, UM members were matched based on their propensity score to the nearest non-UM members using claims-derived cardiac risk scores. These scores reflected the probability of cardiovascular hospitalizations in the post-implementation period for each member and were estimated using predictive variables from the pre-implementation period.
The list of ICD–9 diagnoses codes for cardiovascular conditions was adapted from the SCORE project (Conroy 2003) and defined as International Classification of Diseases, Ninth Revision (ICD–9) diagnoses codes 401 through 414 and 426 through 443 but did not include the following ICD–9 diagnoses codes for definitely non-atherosclerotic causes of hospitalization: 426.7, 429.0, 430.0, 432.1, 437.3, 437.4, and 437.5. Logistic regression was used to calculate the probability of cardiovascular hospitalization; a p-value of 0.05 was used to retain variables considered good predictors of such hospitalizations. The final model included gender, age, history of cardiovascular hospitalization, number of distinct medications, and history of cardiac imaging use as the top five claims-derived predictors with the highest discriminatory power. Using estimated propensity scores for cardiovascular hospitalization, the UM cohort members were matched with the non-UM cohort members who had similarly predicted probability using Greedy nearest neighbor 1:1 ratio matching techniques (Parsons 2012).
Risk group stratification
Recognizing that the need for diagnostic imaging varies with cardiac risk status, AIM gathers clinical data for each health plan member for whom a cardiac imaging test was requested to calculate a cardiac risk score and categorizes members into low, medium, or high-risk groups. These data serve as critical inputs into the clinical review process. Included in AIM’s stratification model are clinical data such as age, diagnosis of diabetes, gender, smoking status, systolic blood pressure, and total cholesterol levels. The availability of such clinically based risk categorization data presented an opportunity to compare our claims-derived risk scores against clinical data on the subset of the UM population that requested cardiac imaging through AIM. To evaluate the performance of the claims-based risk model, average claims-based risk scores (reflecting the probability of cardiac hospitalization) were calculated for patients in each AIM-defined risk group (low, medium, and high) for the subset of members in our study for whom AIM provided clinically based risk group identification. The results of this check showed that the average claims-based risk score (probability of cardiac hospitalization) was 0.0131, 0.0255, and 0.0366 (or 1.31%, 2.55% and 3.66%) for low, medium and high AIM-defined risk subgroups, respectively. The claims-derived risk scores reflected consistently higher risk from low- to high-risk cohorts in AIM’s clinically determined categorization. For the purpose of examining differences in utilization rate changes across the three claims-derived risk groups, members with a greater than 5% percent chance of cardiac hospitalization were classified as high risk (5% being about five times higher than the population average), medium risk was represented by 2% to 5% probability, and low risk was less than 2% probability.
The outcome measured in this study was the difference between the change in echocardiography utilization rate in the matched study population with that of each of the low, medium, and high cardiac risk member subgroups.
Descriptive statistics include the mean, standard deviation (SD), and relative frequencies reported for continuous and categorical data, respectively. Differences in descriptive characteristics between the UM and non-UM groups during the baseline and follow up periods were assessed using Pearson’s chi-square tests for categorical data; Student’s t-tests or non-parametric analysis were used for numeric data whenever appropriate. A difference-in-difference (DID) approach was used to test the difference in echocardiography utilization changes net of pre-UM differences and other covariates. DID was assessed with generalized estimating equations (GEE). Poisson distribution with log link was used to model changes in cardiac imaging utilization. Exponentiated coefficients and corresponding confidence intervals (CI) are presented and are interpreted as incident rate ratios (RR) among groups. In addition, the predicted values from the generalized estimating equations model were used to calculate covariate adjusted mean difference values (expressed as echocardiography tests avoided per 1,000 members) to simplify interpretation of the results. All statistical analyses were conducted with SAS 9.2 version software. Alpha was set at 0.05 for each test.
A total of 2.5 million members were identified in the HIRD; approximately 1.2 million in the non-UM and 1.3 million UM cohorts. After propensity-score matching and selecting preferred provider organization (PPO) enrollees, 0.9 million members were assigned to the UM cohort and 1.1 million to the non-UM cohort. These members were stratified by risk group, as shown in Figure 1.
FIGURE 1 Disposition of the study population
Demographic and clinical characteristics at baseline
The mean age for the UM and non-UM cohorts was approximately 34.5 years, and slightly fewer than half of each group was female, as shown in Table 1. Mean Deyo-Charlson Comorbidity Index (DCI) scores (Deyo 1992) were roughly similar in the UM and non-UM groups, ranging from 0.19–0.21, indicating a low comorbidity burden in the relatively young population. The most common comorbidities were dyslipidemia and hypertension, both of which were just below 15% across all the groups. Coronary heart disease (CHD) was reported at a rate of 2.79% in the overall study population, 2.82% in the non-UM cohort and 2.75% in the UM cohort. The proportion of members who received cardiac procedures such as coronary artery bypass grafting (CABG), percutaneous coronary intervention (PCI) or coronary angioplasty, and cardiac catheterization, among other procedures during the baseline period did not exceed 0.2% in any cohort.
|Table 1 Member demographics and clinical characteristics at baseline (10/01/2008 and 09/30/2009) |
| ||Overall ||UM ||Non-UM ||p value |
|N/Mean ||%/SD ||N/Mean ||%/SD ||N/Mean ||%/SD |
|Number of members ||2,006,373 || ||1,126,074 || ||880,299 || || |
|Mean ||34.53 ||18.67 ||34.35 ||19.13 ||34.77 ||18.05 ||<.0001 |
|Female ||986,101 ||49.15 ||561,869 ||49.90 ||424,232 ||48.19 ||<.0001 |
|Type of health plan |
|PPO ||2,006,373 ||100 ||1,126,074 ||100 ||880,299 ||100 ||<.0001 |
|Deyo-Charlson comorbidity index (DCI) Score |
|Mean score ||0.20 ||0.67 ||0.21 ||0.69 ||0.19 ||0.64 ||<.0001 |
|Heart procedures (HP) ||3,795 ||0.19 ||2,000 ||0.19 ||1,795 ||0.20 ||<.0001 |
|Coronary heart disease (CHD) ||56,007 ||2.79 ||31,810 ||2.82 ||24,197 ||2.75 ||0.001 |
|Congestive heart failure (CHF) ||8,038 ||0.40 ||4,887 ||0.43 ||3,151 ||0.36 ||<.0001 |
|Myocardial infarction (MI) ||6,281 ||0.31 ||3,472 ||0.31 ||2,809 ||0.32 ||0.175 |
|Peripheral vascular disease (PVD) ||6,581 ||0.33 ||4,106 ||0.36 ||2,475 ||0.28 ||<.0001 |
|Chronic obstructive pulmonary disease (COPD) ||11,466 ||0.57 ||6,734 ||0.60 ||4,732 ||0.54 ||<.0001 |
|Cardiac valve disease (CVD) ||24,406 ||1.22 ||13,674 ||1.21 ||10,732 ||1.22 ||0.757 |
|Carotid artery disease (CAD) ||6,232 ||0.31 ||3,681 ||0.33 ||2,551 ||0.29 ||<.0001 |
|Hypertension ||275,533 ||13.73 ||153,338 ||13.62 ||122,195 ||13.88 ||<.0001 |
|Diabetes mellitus ||92,326 ||4.60 ||54,449 ||4.84 ||37,877 ||4.30 ||<.0001 |
|Dyslipidemia ||288,179 ||14.36 ||156,996 ||13.94 ||131,183 ||14.90 ||<.0001 |
|Renal disease ||19,836 ||0.99 ||12,061 ||1.07 ||7,775 ||0.88 ||<.0001 |
|Renal failure ||10,388 ||0.52 ||6,426 ||0.57 ||3,962 ||0.45 ||<.0001 |
A total of 1.8 million members were classified as having low risk for a cardiac event; 976,546 in the non-UM group and 778,693 in the UM cohort. The mean age was approximately 31 years across the cohorts and females made up about 50% of each group. The mean DCI score was 0.09 across cohorts, indicating a lower comorbidity burden and reflecting the risk status of the group. The most common comorbidities were dyslipidemia, ranging from 7.8 percent to 9.0 percent, and hypertension, ranging from 4.3 percent to 5.2 percent, as shown in Table 2.
|Table 2 Member demographics and clinical characteristics at baseline by risk category |
| ||Overall ||Non-UM group ||UM group || |
| ||N/Mean ||%/SD ||N/Mean ||%/SD ||N/Mean ||%/SD ||p value |
|Low risk |
|Number of members ||1,755,239 || ||976,546 || ||778,693 || || |
|Age || || || || || || || |
|Mean ||31.33 ||17.49 ||30.78 ||17.69 ||32.03 ||17.22 ||<.0001 |
|Female ||878,372 ||50.04 ||496,089 ||50.80 ||382,283 ||49.09 ||<.0001 |
|Deyo-Charlson Comorbidity Index (DCI) Score |
|Mean ||0.09 ||0.39 ||0.09 ||0.39 ||0.09 ||0.39 ||0.001 |
|Heart Procedures ||39 ||0.00 ||23 ||0.00 ||16 ||0.00 ||0.796 |
|Comorbidities || || || || || || || |
|Coronary heart disease ||6,254 ||0.36 ||3,255 ||0.33 ||2,999 ||0.39 ||<.0001 |
|Congestive heart failure ||525 ||0.03 ||345 ||0.04 ||180 ||0.02 ||<.0001 |
|Myocardial infarction ||211 ||0.01 ||108 ||0.01 ||103 ||0.01 ||0.193 |
|Cardiac valve disease ||7,982 ||0.45 ||4,208 ||0.43 ||3,774 ||0.48 ||<.0001 |
|Carotid artery disease ||218 ||0.01 ||104 ||0.01 ||114 ||0.01 ||0.018 |
|Hypertension ||83,386 ||4.75 ||42,558 ||4.36 ||40,828 ||5.24 ||<.0001 |
|Diabetes mellitus ||25,058 ||1.43 ||14,294 ||1.46 ||10,764 ||1.38 ||<.0001 |
|Dyslipidemia ||146,764 ||8.36 ||76,589 ||7.84 ||70,175 ||9.01 ||<.0001 |
|Medium risk |
|Number of members ||183,128 || ||107,872 || ||75,256 || || |
|Age || || || || || || || |
|Mean ||55.30 ||7.79 ||55.74 ||8.56 ||54.67 ||6.48 ||<.0001 |
|Female ||81,740 ||44.64 ||49,166 ||45.58 ||32,574 ||43.28 ||<.0001 |
|DCI Score |
|Mean ||0.62 ||1.02 ||0.62 ||1.03 ||0.63 ||1.02 ||0.006 |
|Heart Procedures ||550 ||0.30 ||323 ||0.30 ||227 ||0.30 ||0.932 |
|Comorbidities || || || || || || || |
|Coronary heart disease ||21,765 ||11.89 ||11,897 ||11.03 ||9,868 ||13.11 ||<.0001 |
|Congestive heart failure ||1,744 ||0.95 ||1,065 ||0.99 ||679 ||0.90 ||0.065 |
|Myocardial infarction ||1,536 ||0.84 ||864 ||0.80 ||672 ||0.89 ||0.034 |
|Carotid artery disease ||1,888 ||1.03 ||990 ||0.92 ||898 ||1.19 ||<.0001 |
|Hypertension ||134,416 ||73.40 ||75,920 ||70.38 ||58,496 ||77.73 ||<.0001 |
|Diabetes mellitus ||37,017 ||20.21 ||21,544 ||19.97 ||15,473 ||20.56 ||0.002 |
|Dyslipidemia ||97,612 ||53.30 ||54,739 ||50.74 ||42,873 ||56.97 ||<.0001 |
|High risk |
|Number of members ||68,006 || ||41,656 || ||26,350 || || |
|Mean ||61.20 ||9.07 ||62.48 ||9.70 ||59.18 ||7.55 ||<.0001 |
|Female ||25,989 ||38.22 ||16,614 ||39.88 ||9,375 ||35.58 ||<.0001 |
|DCI Score |
|Mean ||1.82 ||1.70 ||1.85 ||1.72 ||1.79 ||1.66 ||<.0001 |
|Heart Procedures ||3,206 ||4.71 ||1,654 ||4.71 ||1,552 ||5.89 ||<.0001 |
|Coronary heart disease ||27,988 ||41.16 ||16,658 ||39.99 ||11,330 ||43.00 ||<.0001 |
|Congestive heart failure ||5,769 ||8.48 ||3,477 ||8.35 ||2,292 ||8.70 ||0.109 |
|Myocardial infarction ||4,534 ||6.67 ||2,500 ||6.00 ||2,034 ||7.72 ||<.0001 |
|Cardiac valve disease ||9,500 ||13.97 ||5,624 ||13.50 ||3,876 ||14.71 ||<.0001 |
|Carotid artery disease ||4,126 ||6.07 ||2,587 ||6.21 ||1,539 ||5.84 ||0.049 |
|Hypertension ||57,731 ||84.89 ||34,860 ||83.69 ||22,871 ||86.80 ||<.0001 |
|Diabetes mellitus ||30,251 ||44.48 ||18,611 ||44.68 ||11,640 ||44.17 ||0.198 |
|Dyslipidemia ||43,803 ||64.41 ||25,668 ||61.62 ||18,135 ||68.82 ||<.0001 |
This risk category had a total of 183,128 members with 107,872 in the non-UM and 75,256 in the UM categories. The mean age was approximately 55 years and women constituted about 45% of each cohort. The mean DCI score was 0.6 across cohorts reflecting both the medium risk status and low comorbidity burden of the study population. More than 70% of the members had a diagnosis of hypertension, greater than 50% in each group had a dyslipidemia diagnosis, and more than 10% in each group had CHD (Table 2).
A total of 68,006 members were classified as high risk, 41,656 non-UM and 26,350 UM. Members had a mean age of approximately 60 years, and women were approximately 40% of each cohort. The mean DCI scores were higher in this risk category, reflecting a comparatively elevated comorbidity burden: 1.79 for the UM and 1.85 for the non-UM cohorts (Table 2). In excess of 80% of the members were diagnosed with hypertension, while more than 60% and 40% had dyslipidemia and diabetes mellitus, respectively. In addition, CHD was reported for approximately 40% of the members in this cohort. During the baseline period, approximately 5% of the cohort received heart procedures.
Changes in echocardiography utilization rates
Changes in Overall Population
Within the overall study population, the UM group had 29 echocardiography procedures per 1,000 members in the pre-UM period versus 24 per 1,000 members post-UM, a utilization reduction rate of 16.9%. For the non-UM cohort, there were 28.5 procedures per 1,000 members in the pre-management period compared with 28 per 1,000 in the post-management period, a reduction rate of 1.51%, as shown in Table 3. Comparing the post and pre-UM implementation periods, the percentage point difference in unadjusted utilization rates changes between the UM and non-UM cohorts overall was –15.42%. This trend differential resulted in 4.46 avoided tests per 1,000 members for the UM cohort.
|Table 3 Annual echocardiography utilization rate per 1,000 for all members |
|UM ||Non-UM ||UM minus Non-UM ||UM minus expected UM || |
|PRE ||POST ||%Change* ||PRE ||POST ||%Change* ||Percentage point difference† ||Number of tests avoided per 1000‡ ||Total number of tests avoided in UM population |
|n=880,299 ||880,299 || ||1,126,074 ||1,126,074 || || || || |
|28.92 ||24.03 ||–16.93% ||28.47 ||28.04 ||–1.51% ||–15.42% ||–4.46 ||–4321 |
|High risk || || || |
|n=26,350 ||26,350 || ||41,656 ||41,656 || || || || |
|288.425 ||143.947 ||–50.09% ||243.134 ||150.639 ||–38.04% ||–12.05% ||–34.75 ||–917 |
|Medium risk || || || |
|n=75,256 ||75,256 || ||107,872 ||107,872 || || || || |
|99.886 ||67.33 ||–32.59% ||88.485 ||74.968 ||–15.28% ||–17.32% ||–17.3 ||–1302 |
|Low risk || || || |
|n=778,693 ||778,693 || ||976,546 ||976,546 || || || || |
|13.281 ||15.783 ||18.84% ||12.681 ||17.624 ||38.98% ||–20.14% ||–2.68 ||–2102 |
| *Percent change = %Change = (Post — Pre) / Pre |
†Percentage point difference = (%Change UM — %Change non-UM)
‡Number of tests avoided per 1,000 = Number of imagining tests avoided by UM group = (Post of UM — expected Post of UM given non-UM %change)
Changes by risk group
The largest trend differential (percentage point difference) was observed in the low risk group (–20.14%) followed by the medium (–17.32%) and high risk (–12.5%) groups. A total of 2,102 (49% of all avoided tests) echocardiography tests in the low risk group, 1,302 (30% of all avoided tests) in the medium risk group, and 917 (21% of all avoided tests) in the high risk group were avoided (Table 3).
This means that of the 4.46 tests avoided per 1,000 members of the UM population, 2.19 tests were avoided by the low risk group, 1.34 were avoided by medium risk, and 0.94 were avoided by high risk members.
Results from the DID regression estimating the effect of UM program implementation on echocardiography utilization across UM-group and non-UM groups are shown in Table 4.
|Table 4 Echocardiography utilization rates per 1,000 members — adjusted results* |
| ||Parameter estimate ||LCL ||UCL ||P-value |
|Post period ||0.831 ||0.815 ||0.847 ||<.0001 |
|UM group ||0.975 ||0.993 ||0.956 ||0.0073 |
|Post period x UM group† ||0.843 ||0.865 ||0.823 ||<.0001 |
|Age ||1.006 ||1.005 ||1.006 ||<.0001 |
|Female ||1.167 ||1.151 ||1.184 ||<.0001 |
|Deyo-Charlson Index comorbidity score ||1.063 ||1.057 ||1.070 ||<.0001 |
|Low-risk group (claims-derived)‡ ||0.112 ||0.109 ||0.116 ||<.0001 |
|Medium risk group (claims-derived)‡ ||0.491 ||0.480 ||0.502 ||<.0001 |
|Medication count ||1.028 ||1.027 ||1.030 ||<.0001 |
|Covariate adjusted mean difference§ ||–3.05 ||–2.64 ||–3.46 || |
*Analysis conducted with Generalized Estimating Equations using Poisson distribution and log link function. All coefficients are presented in exponentiated form and are interpreted as incident rate ratios.
†Post period x UM group: difference-in-difference coefficient indicates the change in UM group after UM program implementation in comparison to non-UM group and pre-program period.
‡Compared to the high risk group, low and medium risk variables were defined as 1 for presence of low/medium risk and 0 otherwise.
§Covariate-adjusted mean difference is obtained from the results of difference-in-difference modeling. Mean difference indicates the change in echocardiography tests per 1,000 members in UM group after program implementation in comparison to the non-UM group and pre-implementation period.
The primary coefficient of interest in the model was an interaction term (post period X UM group), which estimated the effect of implementing the UM program, while controlling for baseline differences and other model covariates.
The UM-group experienced a significant 15.7% (RR = 0.843, [CI: 0.823, 0.865]) reduction in echocardiography utilization during the post-implementation period when compared with the non-UM group and baseline utilization. Outputs from the GEE modeling (the adjusted reduction estimate and predicted values for utilization in non-UM group) were used to demonstrate that –3.05 [CI: –3.46, –2.64] tests were avoided per 1,000 members in the UM-group, which represents covariate adjusted mean difference between the two cohorts.
Claims-derived cardiac risk has a significant relationship to echocardiography utilization. Compared with the high-risk group, the low risk group had 88.8% (RR = 0.112, [CI: 0.109, 0.116]) lower utilization, and the medium-risk group had 50.9% (RR = 0.491, [CI: 0.480, 0.502]) lower utilization. As expected, age and comorbidity burden (expressed by both DCI score and the number of distinct medications) increased echocardiography utilization. Being female also was associated with slightly increased use of imaging.
Significant decreases were observed in the rate of echocardiography utilization in the UM cohort relative to the non-UM group overall and across the cardiac-risk subgroups. As might be expected, the low and medium risk subgroups contributed the greatest proportion of avoided echocardiography tests. This finding reinforced the hypothesis that while UM programs may result in reductions in overall service volume, they are guided by the principle that health care services should be necessary and appropriate. (Duszak 2012, Otero 2006)
The adjusted rate reduction (15.7 percent ) observed overall in this study was witnessed during a two-year period; there was a one-year implementation (silent) period between the pre- and post-UM periods. This reduction was statistically significant, and it was substantive because it meant that 1 in every 6 echocardiography tests was avoided.
Our findings are consistent with an overall pattern that suggested a flattening in the growth rate of diagnostic imaging utilization in the United States. (Duszak 2012, Lessler 2000, Levin 1996, Levin 2010, Levin 2012, Levin 2012). With respect to echocardiography, such a drop in the growth rate not only reduces proximate costs but also reduces inappropriate procedures which could lead to false positives or inconclusive test results, resulting in additional potentially invasive diagnostic procedures. It is noteworthy that both the UM and non-UM groups experienced service decreases during the study period. This result might be a reflection of the lower trend increase in overall health care spending in the United States between 2009 and 2011. Nonetheless, it is possible that a program spillover effect caused some of the utilization decrease in the non-UM group, as many of the members of the non-UM cohort lived in geographic areas where the AIM program was implemented and used the same provider networks. It is possible that providers have not always differentiated among members who needed prior authorization and those who did not. This possibility may reflect a level of bias; however, its effect would have created favorable results in the non-UM population.
The slowing of growth in utilization of diagnostic imaging marked an important departure from the escalating trends in the last two decades. The fact that a portion of the reductions seen in this study was attributable to UM would suggest its effectiveness and might represent a shift toward appropriate utilization rather than the imposition of unit cost pressures on providers. If true, this result has important implications for managed care companies because targeting high volume procedures is an important focus of cost containment strategies.
The findings in this study are a timely addition to a growing research direction in which population databases are queried for patients with defined characteristics in an effort to quantify usage levels and ascertain the patterns and appropriateness of key aspects of health care utilization. Although the prior authorization requirement for certain imaging procedures has been well established within the commercial insurance industry, the results of this study are timely. Echocardiography, the specific imaging modality evaluated in this study, has not, until recently, been included in prior authorization programs; therefore, the effect of these programs on echocardiography utilization has not been evaluated. A relatively inexpensive and non-invasive procedure, echocardiography has not traditionally been a focus for utilization management. This test, however, can serve as a gateway to additional costlier diagnostics, and the benefits of appropriate management would extend beyond this initial test. This study highlights that significant utilization reductions can be accomplished as more health plans move to manage the use of this procedure.
The population used in this study provided important research advantages, but was subject to the limitations associated with employing administrative claims data for research purposes. Disease severity and other potential confounders that could influence outcomes are generally not observable with claims data and are not typically included in claims-based analyses. While a claims-based risk score was assigned to the population, we did not have clinical information available for the comparison group. Lacking such information means that using clinical elements to compare appropriateness of the tests avoided was out of the scope of this study and could be an area for further work. Because the study sample was from a large managed care database and included continuously enrolled patients, it might not be possible to replicate or generalize these findings across other demographic groups.
The study design, using a DID estimator, assumes that underlying trends in the outcome variables would be similar for both groups, in the absence of a UM program. Unmeasured factors such as changing economic conditions are assumed to affect both groups in similar ways. This assumption is not testable but our careful selection of covariates included in the cardiac risk scoring model and rigorous approach to matching members from both groups makes this a reasonable assumption.
Finally, this study focused on outpatient services and excluded emergency room (ER) procedures. Some members who received denials in the outpatient setting could have been redirected to the ER, however this care pathway was out of the scope of this analysis.
This study demonstrated that UM programs targeting outpatient echocardiography reduced testing rates significantly overall and across the different cardiac risk categories. The fact that utilization avoidance was greatest in the low-risk group suggested that UM could have an important effect on promoting service appropriateness. In addition, UM-derived benefits from avoiding unnecessary additional follow-up procedures may include lower costs and reduced exposure to radiation from follow up testing. It is often difficult to measure the effect of UM programs given the challenges of identifying appropriate comparison groups. This pre-post design with cohorts matched by cardiac risk status demonstrated that there can be a significant effect from claims-compliant UM programs.
Bhatia RS, Carne DM, Picard MH, Weiner RB. Comparison of the 2007 and 2011 Appropriate Use Criteria for Transthoracic Echocardiography in Various Clinical Settings. J Am Soc Echocardiogr. 2012.
Bonow RO, Carabello BA, Chatterjee K, de LA, Jr., Faxon DP, Freed MD, Gaasch WH, Lytle BW, Nishimura RA, O’Gara PT, O’Rourke RA, Otto CM, Shah PM, Shanewise JS. 2008 focused update incorporated into the ACC/AHA 2006 guidelines for the management of patients with valvular heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to revise the 1998 guidelines for the management of patients with valvular heart disease). Endorsed by the Society of Cardiovascular Anesthesiologists, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons. J Am Coll Cardiol. 2008;52:e1–142.
Conroy RM, Pyorala K, Fitzgerald AP, Sans S, Menotti A, De Backer G, De Bacquer D, Ducimetiere P, Jousilahti P, Keil U, Njolstad I, Oganov RG, Thomsen T, Tunstall-Pedoe H, Tverdal A, Wedel H, Whincup P, Wilhelmsen L, Graham IM. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J. 2003;24:987–1003.
Congressional Budget Office. The long-term outlook for healthcare spending. Available at: http://www.cbo.gov/ftpdocs/87xx/doc8758/maintext.3.1.shtml. Accessed Oct. 10, 2012.
Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD–9–CM administrative databases. J Clin Epidemiol. 1992;45:613–619.
Diagnostic Imaging Utilization Management 2012 Program Guidelines V.8.0 Effective Date: November 1, 2012. AIM Specialty Health, Chicago, IL.
Douglas PS, Garcia MJ, Haines DE, Lai WW, Manning WJ, Patel AR, Picard MH, Polk DM, Ragosta M, Ward RP, Weiner RB. ACCF/ASE/AHA/ASNC/HFSA/HRS/SCAI/SCCM/SCCT/SCMR 2011 Appropriate Use Criteria for Echocardiography. A Report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, American Society of Echocardiography, American Heart Association, American Society of Nuclear Cardiology, Heart Failure Society of America, Heart Rhythm Society, Society for Cardiovascular Angiography and Interventions, Society of Critical Care Medicine, Society of Cardiovascular Computed Tomography, and Society for Cardiovascular Magnetic Resonance Endorsed by the American College of Chest Physicians. J Am Coll Cardiol. 2011;57:1126–1166.
Duszak R, Jr., Berlin JW. Utilization Management in Radiology, Part 1: Rationale, History, and Current Status. J Am Coll Radiol. 2012;9:694–699.
Fleisher LA, Beckman JA, Brown KA, Calkins H, Chaikof EL, Fleischmann KE, Freeman WK, Froehlich JB, Kasper EK, Kersten JR, Riegel B, Robb JF, Smith SC, Jr., Jacobs AK, Adams CD, Anderson JL, Antman EM, Buller CE, Creager MA, Ettinger SM, Faxon DP, Fuster V, Halperin JL, Hiratzka LF, Hunt SA, Lytle BW, Nishimura R, Ornato JP, Page RL, Riegel B, Tarkington LG, Yancy CW. ACC/AHA 2007 Guidelines on Perioperative Cardiovascular Evaluation and Care for Noncardiac Surgery: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 2002 Guidelines on Perioperative Cardiovascular Evaluation for Noncardiac Surgery) Developed in Collaboration With the American Society of Echocardiography, American Society of Nuclear Cardiology, Heart Rhythm Society, Society of Cardiovascular Anesthesiologists, Society for Cardiovascular Angiography and Interventions, Society for Vascular Medicine and Biology, and Society for Vascular Surgery. J Am Coll Cardiol. 2007;50:1707–1732.
Hendel RC. Utilization management of cardiovascular imaging pre-certification and appropriateness. JACC Cardiovasc Imaging. 2008;1:241–248.
Iglehart JK. Health insurers and medical-imaging policy—a work in progress. N Engl J Med. 2009;360:1030–1037.
Lessler DS, Wickizer TM. The impact of utilization management on readmissions among patients with cardiovascular disease. Health Serv Res. 2000;34:1315–1329.
Levin DC, Rao VM, Parker L. Trends in the utilization of outpatient advanced imaging after the deficit reduction act. J Am Coll Radiol. 2012;9:27–32.
Levin DC, Intenzo CM, Rao VM, Frangos AJ, Parker L, Sunshine JH. Comparison of recent utilization trends in radionuclide myocardial perfusion imaging among radiologists and cardiologists. J Am Coll Radiol. 2005;2:821–824.
Levin DC, McArdle GH, Lockard CD. Capitated contracting in radiology: negotiating techniques, financial calculations, and utilization management. Radiology. 1996;198:651–656.
Levin DC, Rao VM, Parker L. Trends in Levin DC, Parker L, Halpern EJ, Julsrud PR, Rao VM. The lack of growth in use of coronary CT angiography: is it being appropriately used? AJR Am J Roentgenol. 2011;196:862–867.
Levin DC, Rao VM, Parker L. Physician orders contribute to high-tech imaging slowdown. Health Aff (Millwood ). 2010;29:189–195.
Levin DC, Rao VM, Parker L, Frangos AJ. The sharp reductions in medicare payments for noninvasive diagnostic imaging in recent years: will they satisfy the federal policymakers? J Am Coll Radiol. 2012;9:643–647.
Mansour IN, Razi RR, Bhave NM, Ward RP. Comparison of the Updated 2011 Appropriate Use Criteria for Echocardiography to the Original Criteria for Transthoracic, Transesophageal, and Stress Echocardiography. J Am Soc Echocardiogr. 2012.
Medicare Payment Advisory Commission. A Data Book: Healthcare Spending and the Medicare Program, (MedPAC) 2012. http://www.medpac.gov/documents/Jun12DataBookEntireReport.pdf accessed October 11,
Mieres JH, Shaw LJ, Arai A, Budoff MJ, Flamm SD, Hundley WG, Marwick TH, Mosca L, Patel AR, Quinones MA, Redberg RF, Taubert KA, Taylor AJ, Thomas GS, Wenger NK. Role of noninvasive testing in the clinical evaluation of women with suspected coronary artery disease: Consensus statement from the Cardiac Imaging Committee, Council on Clinical Cardiology, and the Cardiovascular Imaging and Intervention Committee, Council on Cardiovascular Radiology and Intervention, American Heart Association. Circulation. 2005;111:682–696.
Otero HJ, Ondategui-Parra S, Nathanson EM, Erturk SM, Ros PR. Utilization management in radiology: basic concepts and applications. J Am Coll Radiol. 2006;3:351–357.
Parikh PB, Asheld J, Kort S. Does the revised appropriate use criteria for echocardiography represent an improvement over the initial criteria? A comparison between the 2011 and the 2007 appropriateness use criteria for echocardiography. J Am Soc Echocardiogr. 2012;25:228–233.
Parsons, LS. Reducing bias in a propensity score matched-pair sample using greedy matching techniques. In SAS SUGI 26, 2001, Paper 214–26, http://www2.sas.com/proceedings/sugi26/p214-26.pdf , accessed Oct.12, 2012.
Warnes CA, Williams RG, Bashore TM, Child JS, Connolly HM, Dearani JA, del Nido P, Fasules JW, Graham TP, Jr., Hijazi ZM, Hunt SA, King ME, Landzberg MJ, Miner PD, Radford MJ, Walsh EP, Webb GD, Smith SC, Jr., Jacobs AK, Adams CD, Anderson JL, Antman EM, Buller CE, Creager MA, Ettinger SM, Halperin JL, Hunt SA, Krumholz HM, Kushner FG, Lytle BW, Nishimura RA, Page RL, Riegel B, Tarkington LG, Yancy CW. ACC/AHA 2008 guidelines for the management of adults with congenital heart disease: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Develop Guidelines on the Management of Adults With Congenital Heart Disease). Developed in Collaboration With the American Society of Echocardiography, Heart Rhythm Society, International Society for Adult Congenital Heart Disease, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons. J Am Coll Cardiol. 2008;52:e143–e263.
Wickizer TM, Lessler D. Utilization management: issues, effects, and future prospects. Annu Rev Public Health. 2002;23:233-254.
Wickizer TM. Controlling outpatient medical equipment costs through utilization management. Med Care. 1995;33:383–391.
Weiner SN, Komarow M. Utilization management and noninvasive diagnostic imaging. Manag Care Interface. 2005;18:34–38.
Andrea DeVries, PhD
Acknowledgements: Bernard B. Tulsi, MSc, provided writing and other editorial support for this manuscript.
Funding source: This research project was sponsored by AIM Specialty Health, a specialty benefit manager.
Conflict disclosure: A De V, GS, AA, and BBT disclose that they are employees of HealthCore, a research subsidiary of WellPoint, serving as paid consultants to AIM in the development and execution of this manuscript and the underlying study. TP discloses that he is an employee of AIM, which provided funding for this study.