 To evaluate the prevalence of previously unrecognized osteoporosis in men admitted for long-term rehabilitation nursing home care.

A total of 1179 consecutive admissions to a VA rehabilitation center were reviewed. Men who were already diagnosed with osteoporosis, had confounding medical illness, were unable to complete the study, or who declined to participate were excluded. A total of 153 patients were enrolled and 106 were evaluated. Measurements included dual-energy X-ray absorptiometry of the hip and lumbar spine, biochemical and hormonal studies, and functional evaluation.

A total of 33 patients (31.1%) had osteoporosis (T-score at any site lumbar spine, total hip, or femoral neck <−2.5). Patients with osteoporosis were significantly older than those without: 68.4 ± 13.2 years versus 62.7 ± 12.1 years (P <.05), respectively. Body mass index (BMI) and weight were lower in men with osteoporosis: 23.4 ± 3.9 kg/m2 versus 28.7 ± 7.08 kg/m2 and 72.6 ± 14.4 kg versus 90.3 ± 23.8 kg, respectively (both, P <.001). There were no differences in use of medications thought to affect bone metabolism or functional status, or in hormonal and metabolic measurements. Hip and spine bone density were correlated (r = 0.3, P <.05). Multivariate analysis showed that hip bone density was independently associated with BMI.

Hip osteoporosis is common in this unscreened population, suggesting that screening should be more widely performed in veterans admitted to rehabilitation units. These data suggest that nutritional status could impact osteoporosis risk.

The prevalence of previously unrecognized osteoporosis in men admitted for long-term rehabilitation nursing home care was evaluated in a cross-sectional study.

Of the men admitted to a VA rehabilitation unit, 31.1% had osteoporosis. 

Men with osteoporosis were older and thinner than those without osteoporosis, but both groups had similar hormonal measurements, estimates of nutritional status, and functional status.

Because of the substantial morbidity and costs associated with osteoporosis in men, as well as our failure to identify simple predictors, it is important to emphasize screening programs to identify men at risk and institute appropriate interventions.

Osteoporosis is a significant health problem in the United States, and its impact is growing as the population ages. The annual financial burden from osteoporosis has been estimated at $13.9 billion (in 1995), primarily reflecting the cost of hip fracture, surgical repair, nursing home stays, and rehabilitation.

 A similar impact has been reported in other industrialized countries such as Belgium,

 and this impact will be felt throughout the world.

 To date, most of the studies on osteoporosis have focused on women, but it is increasingly apparent that men are affected as well. Estimates of osteoporosis in men have varied, but available data suggest that up to 30% of elderly (age >65 years) men will have an osteoporotic fracture in their lifetime.

 Available data on osteoporosis in nursing home patients are sparse, especially in men, but published estimates of prevalence of subnormal bone density range between 40% and 60%.

 Given that many high-risk patients (eg, patients on chronic glucocorticoids

) are not being screened in a systematic manner, available screening and treatment data may underestimate the total prevalence of osteoporosis in our population.

 Nursing home patients are at significant risk from osteoporosis and its sequelae, due in part to immobility, concurrent medical problems, poor nutrition, and especially the risk of falls.

 The potential benefits of exercise and hip pads to reduce the risk of hip fracture have been studied.

 We evaluated the prevalence of osteoporosis in a high-risk population: the largely male, elderly veteran population at the Center for Rehabilitation and Extended Care (CREC) of the VA Northern California Health Care System. In addition, we estimated factors that may contribute to metabolic bone disease.

 The CREC currently houses up to 120 patients on 3 wards. These patients are mostly male and elderly. Since January 1999, we have been using an on-site dual-energy x-ray absorptiometry (DEXA) machine to evaluate both trabecular and cortical bone density in the lumbar spine and hip.

 Consecutive male patients admitted to the CREC were recruited for the study. We excluded “short-stay” individuals (eg, those recovering from procedures and admitted for <23 hours). We also excluded patients admitted for respite care, those with an inability to cooperate because of dementia or other conditions, those with cancer metastatic to bone or known metabolic bone disease, and patients previously evaluated, diagnosed, or treated for osteoporosis. We obtained approval to approach the patients from providers caring for them; this step also gave us an opportunity to confirm, for example, exclusion for dementia. Patients also were approached in person; if they were off the ward for a home visit, physical therapy, and so forth, they might have been excluded. Written informed consent was obtained from each patient or his legal guardian. The study was approved by the Human Studies Subcommittee of the VA Northern California Health Care System.

 Screening was performed by DEXA, using an on-site Lunar XRC-1 (GE-Lunar, Madison, WI), which results in modest radiation exposure and allows serial measurements and evaluation of therapy.

 This instrument has a precision of 2% to 3% on repeated measurements of bone density. Osteoporosis was defined by World Health Organization criteria using T-scores of -2.5 or greater at either the lumbar spine, total hip, or the femoral neck. Bone density data were stratified by race-specific criteria in the Lunar database, in keeping with station policy.

 included overall nutritional health estimated by complete blood count (providing data on white blood cell count and hematocrit) and routine chemistries, including albumin and 25-hydroxy vitamin D. To estimate calcium absorption, 24-hour urinary calcium was used. The prostate-specific antigen and serum protein electrophoresis tests were used to estimate the possibility of occult malignancy contributing to poor bone density. Bone turnover was estimated by total alkaline phosphatase. Bone-active hormones were estimated by measurement of testosterone, parathyroid hormone (PTH), thyroid-stimulating hormone (TSH), and free thyroxine (to assess the thyroid axis40).

 Because of the variable and occasionally prolonged in-facility residence, laboratory data closest to the DEXA scan were selected if an analyte was measured more than once. Assayed but undetectable prostate-specific antigen was entered as “0” in the analysis. Parathyroid hormone that was assayed but not detected was entered as one-half the lower limit of detection in the analysis.

 Overall health was estimated in part by documenting active medical problems. We also calculated functional status both through case mix index (CMI) and activities of daily living (ADL) scores through the minimum data set package.

 Data are expressed as mean ± SD. The t test was used to compare continuous characteristics of the populations diagnosed with and without osteoporosis. The χ2 test was used to compare discontinuous variables such as ethnicity and use of bone-active medications such as phenytoin, barbiturates, and systemic glucocorticoids. Multivariate analysis using multiple linear regression (SigmaStat version 3.0; Systat Software, Inc, Point Richmond, CA) was used to identify predictors of bone density from variables including age, comorbidity (number of active diagnoses, functional status), nutrition (albumin, hematocrit, white blood cell count, vitamin D, and urinary calcium), and bone-active hormones (testosterone, PTH, TSH, and thyroxine). These variables were adapted from widely accepted causes of secondary osteoporosis in adults.

 The sponsor had no role in the collection of data, its analysis or interpretation, or approval of the manuscript.

 Between July 2000 and April 2006, 1179 consecutive admissions of men were reviewed. A total of 116 patients (9.84%) had already been screened or were receiving therapy for osteoporosis. Of the remaining 1063, 113 men (10.6%) were considered to be too demented on initial chart review to participate, 93 (8.7%) had cancer metastatic to bone or were receiving end-of-life care, and 17 (1.6%) died or were discharged prior to the study. The providers for all other patients were contacted to ascertain whether they were capable of participating. A total of 687 men were approached and either declined to participate, participation was discouraged by their provider, or participation was impractical for logistical reasons during the patient’s inpatient stay. Of the remainder, 153 patients met study criteria and gave written consent; 106 completed the studies. A total of 33 men (31.1%) were diagnosed with osteoporosis in any site; 26 had hip only, 4 both hip and spine, and 2 had spine only. One man had osteoporosis at Ward’s triangle only, but he was not included in the analysis. A total of 42 men had osteopenia (T-scores of −1 to −2.5 at lumbar spine or hip sites), 32 men had no evidence of osteoporosis or osteopenia, and 46 men enrolled in but did not complete the study, usually because of being discharged from the facility prior to completion. Of the 106 men who completed the study, 66 (62.3%) were self-described as white, 33 (31.1%) as African American, 6 (5.7%) as Hispanic, and 1 (0.9%) as Asian. This proportion is similar to the veteran population in our catchment area.

 The study was suspended in May 2002 due to a suspension of the facility Federalwide Assurance (FWA); with the restoration of the FWA, the study recommenced in spring 2003. This hiatus allowed us to estimate changes in practice patterns in osteoporosis screening. Prehiatus, 42 of 565 admitted patients (7.43%) had been screened or were treated; posthiatus, 74 of 614 patients (12%) had been screened or were treated. This difference in rates was statistically significant by the χ2 test.

 Results comparing men with and without osteoporosis are shown in the 

. The men with osteoporosis were slightly, but significantly, older than those without osteoporosis. Other clinical markers, including 24-hour urinary calcium, PTH, and alkaline phosphatase, were similar in men with and without osteoporosis. Both groups had low-normal 25-hydroxy vitamin D and low urinary calcium. Thyroidstimulating hormone was slightly but not significantly lower in men with osteoporosis compared with men without osteoporosis, although generally in the normal range. Weight and body mass index (BMI) were both significantly lower in men with osteoporosis than in those without.

 Comorbidity—as manifested by length of stay, number of active diagnoses on the problem list, ADL scores, and CMI scores&mdash;was comparable between the 2 groups. The number of men receiving medications thought to have bone activity, specifically barbiturate, phenytoin, and systemic glucocorticoids, was too small to calculate significant differences between the groups of men with and without osteoporosis. (Among men identified with osteoporosis, 1 man was on phenobarbital and 1 man was on ongoing glucocorticoids; none were on Dilantin. Among men without osteoporosis, none were taking phenobarbital or Dilantin, and 4 were taking glucocorticoids.) The men who enrolled in but did not complete the study were not significantly different from those with osteoporosis (data not shown) in terms of age and indices of comorbidity. The data were too incomplete otherwise to permit a more thorough comparison.

 Because only 153 of 840 men who were eligible and approached for the study agreed to participate, we were concerned that this sample be representative of the CREC population over the time frame of the study. Thus, we reviewed the roster of men deemed eligible for the study and identified a sample: All those whose Social Security number ended in “7” had their data reviewed (number chosen by a random number generator). This “unstudied” group (n = 63) had a self-described ethnic distribution similar to that of the study population. The “unstudied” group was slightly, but significantly, older (70.5 ± 11.7 years vs 64.5 ± 12.7 years; P <.01), but otherwise was similar in terms of active diagnoses, functional status, weight, and BMI.

 There was no difference in ethnic composition between men with and without osteoporosis by the χ2 test. Multiple linear regression analysis showed that the significant predictors of hip bone density were lumbar spine bone density (P <.05) and BMI (P <.001). Although lumbar spine bone density predicted hip density, very few men had overt osteoporosis of the spine. Because degenerative changes in the spine can contribute to spuriously high bone density readings, we evaluated our osteoporosis patients for spine degeneration. Lumbar DEXA readings were reviewed, and any patient who was described as having scoliosis, compression, spurring, degeneration, arthritis, or deformity was labeled. When these “degenerative” patients (n = 23) were compared with osteoporotic patients without degenerative changes (n = 10), there was a significant difference in lumbar T-score (−2.19 ± 0.7 compared with −0.51 ± 1.8, respectively; P <.001). Age, biochemical measurements, ADL scores, and CMI scores did not contribute significantly. There was no seasonal variation in 25-hydroxy vitamin D (not shown).

illustrates the correlation of hip bone density and spine bone density, as measured by T-score.

 In this study, we found that approximately 30% of a sample of men admitted to a VA rehabilitation unit had osteoporosis, and despite being older and thinner than a control group, they had no demonstrable metabolic, functional, or ethnic differences. In a very broad sense, our data are consistent with those of other recent studies. A study of men from the Miami VA nursing home demonstrated that slightly more than 62% had low bone mass (~36% with osteopenia, ~27% with osteoporosis), obesity seemed to correlate with osteoporosis (unlike our study), and ambulatory status did not predict bone mass.

 Although the overall prevalence of osteoporosis in the study by Paniagua et al was comparable to that reported here, their study was unlike ours in other ways

: they used heel densitometry, exclusion criteria were not defined, and demented patients were included and indeed made up a majority of the population. In another study that used heel densitometry as a screening technique, Gloth and Simonson evaluated more than 30,000 skilled nursing facility residents in 26 states; ~42% had T-scores below −2.5. Bedbound patients, those being treated for osteoporosis, and those with a “short life expectancy” were excluded.

 Although these reported prevalence statistics for osteoporosis were comparable to those in our study, our project reported vitamin D, testosterone, TSH, and PTH data, which these earlier reports did not. In a UK study, Aspray et al reported that nursing home residents have an osteoporosis prevalence of 69%; however, the majority of their patients were women.

 A study from New York suggested that elderly men and women admitted to an urban tertiary care hospital with osteoporotic fractures were a heterogeneous group, without clearly consistent predictors.

 In recognition of the enormous morbidity, mortality, and financial impact of osteoporosis, Tucci emphasized the need to screen and treat aggressively to prevent future fractures,

 particularly since osteoporotic fractures in men carry greater morbidity and mortality than those in women.

 Our data provide indirect evidence in support of the need to screen veterans admitted to rehabilitation facilities aggressively, because we were so limited in our ability to identify key factors that might predispose to osteoporosis in men. A randomized controlled trial comparing screening and not screening with respect to fracture outcome might be of value. Similarly, a study demonstrating that treatment prevents fracture in this population would be of interest. It also may be that enough data have accrued in a similar population to allow a modeling study.

 Inadequate calcium absorption and vitamin D may contribute to osteoporosis in a nursing home population. Although there was no clear difference in these parameters for patients with and without osteoporosis, both calcium absorption (as measured by 24-hour urine calcium) and circulating vitamin D were at the lower end of normal. Recent studies

 have suggested that the optimal circulating vitamin D concentration is 30 ng/mL or above; both our study populations were well below that level. This may be related to an age-associated decline in skin synthesis

 and appears to transcend age and race distinctions.50 Our assay did not distinguish between 25-hydroxy vitamin D3 and 25-hydroxy vitamin D2. The importance of nutrition may be indicated by our observation that hip bone mass density (T-score) was significantly related to BMI and body weight. Although bone mass is a component of lean body mass, BMI may reflect global nutrition as well as ongoing mechanical bone loading. These findings are distinct from those of Paniagua et al, who reported an increase in osteoporosis in heavier patients,

 suggesting that factors other than BMI alone (eg, activity level) play an important role.

Articles

CLINICAL

Men in a VA rehabilitation unit who had osteoporosis were older and thinner, but otherwise similar (metabolic and functional status) to control subjects.

Prevalence of Osteoporosis in Men in a VA Rehabilitation Center

 To investigate whether compliance during the first 2 years of statin therapy is associated with reduced hospitalization rates and direct medical costs during year 3.

An integrated pharmacy and medical claims database was used to identify adult patients with a new statin prescription between July 1, 2001, and June 30, 2002. The study tracked statin prescription refills during the first 2 years after the initial statin claim and tracked hospitalizations and direct medical costs during the first 3 years.

Patients were stratified according to compliance in the first 2 years using the medication possession ratio, where 80% or higher is compliant and less than 80% is noncompliant. The relationship between compliance rates and direct medical costs was evaluated using a generalized linear model. Adjusting for covariates that may affect cardiovascular risk, the relationship between compliance and the likelihood of hospitalization was assessed using logistic regression models.

The 2-year medication possession ratio was 80% or higher in 3512 patients (compliant) and was less than 80% in 6715 patients (noncompliant). Compared with the noncompliant patients, the compliant patients during year 3 had significantly fewer hospitalizations (16% vs 19%) and lower total direct medical costs (excluding the cost of statin therapy) ($4040 vs $4908 per patient) (P <.01 for both).

Compliance with statin therapy in the first 2 years of use may reduce hospitalization rates and direct medical costs in the subsequent year.

	Compliance rates of 80% or higher during the first 2 years of statin therapy were associated with significantly reduced hospitalization rates and direct medical costs in the subsequent year among a heterogeneous group of statin users. 

		Both all-cause and cardiovascular-related hospitalizations were significantly reduced in patients with compliance of 80% or higher versus less than 80%.

		Total direct medical costs were lower in the group with compliance of 80% or higher versus less than 80%.

		The greatest potential incremental benefit can be achieved by motivating patients who are 60% to 79% compliant to be 80% compliant or higher.

An estimated 16 million Americans have coronary heart disease (CHD).

 It is the most common cause of mortality among American adults. Because of the morbidity and mortality burden of CHD, it is not surprising that costs associated with this condition are high. In the United States in 2008, direct medical costs for CHD exceeded $87 billion, while indirect costs (lost productivity of the patient and the caregiver) accounted for an additional $70 billion.

 Therefore, identifying strategies for reducing the incidence of CHD can lead to significant monetary savings.

One proven strategy for prevention of CHD is to reduce low-density lipoprotein cholesterol (LDL-C) levels.

 A considerable body of evidence demonstrates that hydroxymethyl glutaryl coenzyme A reductase inhibitors (statins) are effective and well-tolerated agents for decreasing LDL-C levels and for reducing cardiovascular (CV) events.

However, treatment can be expected to provide a clinical benefit only  in patients who take their medication continuously on a long-term basis. Several studies have demonstrated that LDL-C level and CV risk reduction with lipid-lowering therapy is greater in patients who are compliant with therapy

 and that poor compliance may account for the difference in outcomes between randomized controlled trials and clinical practice.

 Data are limited about the effect of compliance on outcomes within the first 1 to 2 years of treatment with statins. Veterans Healthcare Administration data indicate that compliance during the first 1 to 2 years of statin treatment can reduce mortality in the subsequent year among patients with diabetes.

 However, that study was conducted in a high-risk population, and it is unknown whether early compliance with statin therapy can provide tangible benefits in terms of healthcare resource utilization among a more heterogeneous population.

The objective of the present study was to evaluate the effect of statin therapy compliance during the first 2 years of treatment on healthcare resource utilization in year 3 from the payer perspective. We hypothesized that patients with continuous statin treatment compliance will incur lower healthcare costs in the following year than patients with less compliance, even in early stages of treatment.

This was a retrospective analysis of integrated medical and pharmacy claims data between January 1, 2000, and June 30, 2004. Adult patients who met the following criteria were identified and included in the study: (1) received a statin prescription between July 1, 2001, and June 30, 2002, and had no prescription for statins in the 6 months before this prescription (ie, they were presumed to be new to statin therapy); (2) had continuous coverage by the health plan for the 6 months before the initial (index) statin prescription and 3 years after the index statin prescription; (3) filled 2 prescriptions on 2 different dates or 1 prescription with at least 45 days supply in the first year following the index date; (4) were at least 18 years old at the index date; and (5) underwent no pregnancy or fertility procedures during the follow-up period. All patients had to be continuously enrolled in the health plan to be eligible for inclusion; therefore, no study patients died during the follow-up period, and the effect of statin compliance on mortality was not examined.

Medical claims were obtained from a deidentified administrative claims database that is available for research purposes. The database was maintained by a health plan organization for all medical plan types, including health maintenance organizations, preferred provider organizations, traditional fee-for-service plans, and Medicare Advantage. Drug utilization data were obtained from the prescription claims database of Medco Health Solutions, Inc, the pharmacy benefits management company responsible for the prescription benefit plan among this population.

For each patient (using member identification), the following medical claim information was collected: service date, place of service, copayment amount, deductible amount, amount the health plan paid, 

International Classification of Diseases, Ninth Revision, Clinical Modification 

) diagnosis code. Similarly, the following pharmacy claim information was collected for each patient’s prescription history: date filled, days supply (number of days), National Drug Code number, drug name, copayment amount, deductible amount, and the amount the health plan paid. All pharmacy claims, including statins, were included in the analysis of drug utilization. Data on age and sex were drawn from an eligibility database maintained by the health plan organization. Medical and pharmacy claims data were integrated at the patient level, with all personal identifying information removed.

Medical and pharmacy claims data were analyzed for the 6 months before the first statin prescription (index date) and for the 3 years after the index prescription. Drug and medical claims were tracked concurrently for the 3-year period after the index date of initial statin prescription. All statin prescriptions for an individual patient were considered during this period; a switch from one statin to another was considered continuous statin therapy. Because statin costs were based on pharmacy claims, patients with lower compliance incurred lower statin costs, and vice versa.

Medication compliance over the first 2 years of statin therapy was assessed using the medication possession ratio (MPR), defined as the proportion of a period in which a patient had medication supply. The MPR numerator was derived by summing the number of days with prescription medication from the date of the index claim until 2 years after this date. This was then divided by 730 (total number of days in 2 years) to derive the MPR. Patients with an MPR of 80% or higher were considered compliant, and patients with an MPR of less than 80% were considered noncompliant. The MPR was chosen for this analysis because it is the most frequently cited measure of compliance in the literature,

 and 80% is the most common threshold for compliance.

 In a secondary analysis, patients in the sample were stratified into the following 5 categories according to their MPR: 0% to 19%, 20% to 39%, 40% to 59%, 60% to 79%, and 80% or higher.

Using the medical claims data, outcomes in the third year after the index statin prescription were compared between patients who were considered compliant versus noncompliant in the preceding 2 years. The outcomes evaluated during that year were the number of hospitalizations per 100 patients, the percentage of patients hospitalized, and the total medical costs (with and without statin therapy cost) per patient. Data were derived from primary, secondary, and tertiary diagnoses defined by ICD-9-CM codes in medical claims data and from CPT codes for angina, CV disorders, cerebrovascular disorders, and peripheral vascular disease procedures (eAppendix available at www.ajmc.com). In the secondary analysis, the same variables were evaluated in the groups stratified by 20% increments of MPR. The ICD-9-CM and CPT codes were chosen because clinical trials demonstrated an effect of statin therapy on these outcomes

or on reducing morbidity and mortality after these procedures or events.

 All healthcare utilization costs related to each ICD-9-CM or CPT code were calculated from the claims data for each patient during year 3 of statin treatment, including outpatient visits, emergency department (ED) visits, inpatient overnight stays in a hospital, surgery, physical therapy, laboratory, and pharmacy. All statin costs included the cost to the insurer and the patient copayment. An ED visit was defined as a visit to an ED that did not exceed 1 day. If an ED visit progressed to an overnight stay, it was considered a hospitalization.

Baseline demographic characteristics in the compliant and noncompliant groups were compared using χ2 test and t test. A generalized linear model was used for continuous dependent variables (cost), and logistic regression models were used to evaluate the relationship between medication compliance and dichotomous dependent variables (the likelihood of having an event). Statistical significance was assessed using F test for continuous variables (generalized linear models) and Wald χ2 in the logistic regression models, and 95% confidence intervals (CIs) were calculated. The following covariates were used in the logistic regression model: (1) age; (2) sex; (3) chronic disease score; (4) the use of concomitant medications that may increase or decrease the risk of CV events (specifically antiplatelet agents and antiarrhythmics); (5) comorbidities (diabetes, renal disease, congestive heart failure, chronic obstructive pulmonary disease, depression, or hypertension) during the 2 years after the index statin prescription (period of compliance assessment); and (6) CV disorders (based on ICD-9-CM codes) or the need for CV or peripheral vascular procedures (based on procedural claim codes) during the 2 years after the index statin prescription.

Our analysis identified 10,227 patients who met the inclusion criteria. Patients were aged 18 to 93 years (mean age, 58.5 years). Of these, 3512 (34%) had a 2-year MPR of 80% or higher and comprised the compliant group. The other 6715 patients (66%) had a 2-year MPR of less than 80% and comprised the noncompliant group. The mean (SD) MPRs were 92.4% (5.8%) in the compliant group and 40.2% (22.4%) in the noncompliant group.

Patient demographic characteristics at the time of the index statin prescription are given in 

. The compliant group was significantly younger, included fewer female patients, had a higher mean chronic disease score, and had more comorbidities than the noncompliant group (P <.001 for all). Compared with the noncompliant group, the compliant group also included more patients who underwent a CV procedure in the 2 years following the index statin prescription (P = .03) and who had depression or hypertension (P <.001 for both) but included fewer patients with renal disease (P =.04), congestive heart failure (P = .001), and chronic obstructive pulmonary disease (P = .002). There was no significant difference between the 2 groups in the use of antiarrhythmic or antiplatelet therapy.

In the logistic regression model, the adjusted percentage of patients hospitalized during year 3 of statin use was significantly lower in the compliant group than in the noncompliant group (16% vs 19%, P <.01) (

). In addition, patients who were compliant in the preceding 2 years had significantly fewer hospitalizations in year 3 (25 vs 33 per 100 patients, P <.01).

Results of the generalized linear model for per-patient medical costs are shown in 

. Excluding the cost of statin therapy, the adjusted per-patient total medical costs in the third year of statin therapy were significantly lower in the compliant group than in the noncompliant group ($4040 [95% CI, $3601-$4478] vs $4908 [95% CI, $4594-$5222], P <.01). When the cost of statin therapy was included, the adjusted total medical costs per patient were still significantly lower in the compliant group ($4909 [95% CI, $4470-$5347] vs $5290 [95% CI, $4976-$5604], P <.01).

When the data were evaluated by 20% increments of MPR, a similar pattern emerged. The adjusted percentage of patients hospitalized and number of hospitalizations per 100 patients were lower for the group with an MPR of 80% or higher versus groups with lower MPRs (P <.001 for variation across quintiles) (

), as were the total medical costs (including and excluding the cost of statin therapy) (P <.001) (

). The most hospitalizations and the highest costs were seen in the group with the poorest compliance (0%-19% MPR). The incremental difference in adjusted healthcare resource utilization and total medical costs between adjacent categories of MPRs was greatest between groups with MPRs of 60% to 79% and MPRs of 80% or higher. The total per-patient medical costs, excluding the cost of statin therapy, were $4691 (95% CI, $4076-$5306) in the group with MPRs of 60% to 79% and $4035 (95% CI, $3597-$4473) in the group with MPRs of 80% or higher.

Our data indicate that sustained compliance with statin therapy (>80% MPR) in the first 2 years after initiation is associated with reductions in healthcare utilization and direct medical costs in the following year. Using MPRs,  Zhang et al

 showed that early compliance with statin therapy (in the first 1-2 years of use) among a population with diabetes resulted in reduced risk of mortality. Patients with diabetes represent a high-risk group, and similar reductions in mortality or major CV outcomes from early statin compliance in a lower-risk cohort have not previously been demonstrated to our knowledge. Even in a heterogeneous population of statin users, our data indicate that early compliance is associated with fewer all-cause and CV-related hospitalizations and with lower healthcare utilization costs. Unlike Zhang et al, we did not analyze the effect of statin compliance on mortality because our study included only patients who were alive for 3 years after the index statin prescription.

REVIEW ARTICLE

Statin therapy compliance of 80% or higher during the first 2 years of treatment is associated with reduced healthcare resource utilization in the following year.

Is There a Relationship Between Early Statin Compliance and a Reduction in Healthcare Utilization?

To share outcomes and lessons learned from an evaluation of disease management (DM) programs for asthma, congestive heart failure (CHF), and diabetes for TRICARE patients.

Multiyear evaluation of participants in voluntary, opt-out DM programs. Patientcentered programs, administered by 3 regional contractors, provide phone-based consultations with a care manager, educational materials, and newsletters. The study sample consisted of 23,793 asthma, 4092 CHF, and 29,604 diabetes patients with at least 6 months' tenure in the program.

Medical claims were analyzed to quantify program effect on healthcare utilization, medical costs, and clinical outcomes. Multivariate regression analysis with an historical control group was used to predict patient outcomes in the absence of DM. The difference between actual and predicted DM patient outcomes was attributed to the program. A patient survey collected data on program satisfaction and perceived usefulness of program information and services.

Modest improvements in patient outcomes included reduced inpatient days and medical costs, and (with few exceptions) increased percentages of patients receiving appropriate medications and tests. Annual per patient reductions in medical costs were $453, $371, and $783 for asthma, CHF, and diabetes program participants, respectively. The estimated return on investment was $1.26 per $1.00 spent on DM services.

Findings suggest that the DM programs more than pay for themselves, in addition to improving patient health and quality of life. Lessons learned in program design, implementation, effectiveness, and evaluation may benefit employers contemplating DM, DM providers, and evaluators of DM programs.

A multiyear evaluation of 3 TRICARE disease management (DM) programs for asthma, congestive heart failure, and diabetes looked at patient outcomes along 4 dimensions: utilization, financial, humanistic, and clinical.

For all 3 diseases, outcome measures moved in the direction anticipated -- that is, less emergency use and inpatient care, lower medical costs, and (with few exceptions) increased percentages of patients receiving appropriate medications and tests. 

The estimated return on investment was $1.26 per $1.00 spent on DM services. 

These DM programs more than pay for themselves, in addition to improving patient health and quality of life.

Providing patients with information to manage their medical conditions and encouraging positive change in health-related behavior can lead to improved health and avoidance of costly medical expenditures. The Disease Management Association of America identified asthma, congestive heart failure (CHF), and diabetes among the chronic diseases with the greatest potential for disease management (DM).

 There is no consensus, however, on whether DM provides the anticipated benefits. Some studies report positive return on investment (ROI),

 whereas others report negative ROI or inconclusive results.

 The Military Health System, through the TRICARE program administered by the TRICARE Management Activity (TMA), provides care to 9.2 million beneficiaries at an annual cost of $44.8 billion (FY 2009 budget).

 Among current beneficiaries under age 65 years residing in the United States, analysis of administrative records suggests that approximately 80,000 have asthma, 11,000 have CHF, and 225,000 have diabetes.

 TRICARE Management Activity contracted with its 3 regional managed care support contractors (MCSCs) to provide DM services to high-utilization patients with asthma and CHF beginning in September 2006, adding diabetes in June 2007. Although this voluntary, opt-out DM program is ongoing, the findings presented here are based on analysis of patients contacted through September 30, 2008, who had at least 6 months’ tenure in the program. The 6-month-tenure criterion was to allow sufficient time for the MCSCs to provide DM services. The September 2008 cutoff was to ensure 100% completeness of medical claims. The analysis sample consisted of 23,793 asthma, 4092 CHF, and 29,604 diabetes patients.

 TRICARE Management Activity identifies candidates for DM based on high utilization of disease-related services (emergency and outpatient visits, hospitalizations) and, for asthma, high use of short-acting rescue prescriptions. Patient names are forwarded to the MCSCs to verify eligibility and provide DM services. Some patients choose to fully participate and receive personalized telephonic counseling and educational mailings (referred to as “engaged” patients), whereas others decline personalized care but receive newsletters. Some patients opt out of the program.

 The DM content and services follow industry best practices, including (1) an initial 40- to 50-minute baseline assessment by telephone with a care manager; (2) monthly follow-up telephone calls to set goals, assess progress toward those goals, and educate patients about their conditions and self-management; (3) educational materials specific to the patient’s needs (eg, pamphlets, videos, cookbooks); and (4) newsletters and Internet-based materials. All services are educational and are intended to provide patients with disease-specific knowledge and self-management skills. Two MCSCs consider a patient to have “graduated” 12 months after receiving the baseline assessment, while the third considers a patient to have graduated when goals regarding disease understanding and self-management have been met. After graduation, patients continue to receive newsletters and biannual phone contact.

 Best practices for DM evaluation suggest measuring program effect along 4 dimensions: utilization, financial, clinical, and humanistic.

 Using an encrypted patient identifier, we electronically linked patient DM administrative records, medical claims, and a patient survey. Study protocols were approved by an institutional review board.

 Patient outcome metrics were selected using recommendations of the 2nd Annual Disease Management Outcomes Summit.

 Utilization metrics (emergency visits and hospital inpatient days) and financial metrics (medical expenditures) for asthma and CHF are disease specific. Claims are counted only when the disease is listed in the first or second diagnosis code position (first diagnosis position only for asthma-related hospitalization and emergency visits).

 Only a small portion of diabetes-related medical costs are directly for treatment of diabetes; the majority are associated with complications of diabetes (eg, neurologic symptoms, peripheral vascular disease, cardiovascular disease, renal complications, endocrine complications, ophthalmic complications) and higher use of medical care for general medical conditions.

 Therefore, for diabetes we report disease-related utilization and cost but focus on total utilization and cost (with claims for injury, pregnancy, congenital abnormalities, and malignant cancer excluded). All medical costs were adjusted to 2008 dollars using the medical component of the consumer price index.

 Clinical metrics were limited to those available in electronic records and include the percentage of patients receiving an exam or a pharmaceutical in the past year. Asthma metrics were (1) use of long-term controller medications (measured as the percentage of patients with at least 1 dispensed prescription for inhaled corticosteroids, nedocromil, cromolyn sodium, leukotriene modifiers, or methylxanthines); and (2) the percentage of patients with a documented spirometry test. For CHF we determined the percentage of patients with angiotensin-converting enzyme (ACE) inhibitor and beta-blocker prescription fills. Diabetes metrics were (1) the percentage of patients with at least 1 glycosylated hemoglobin (A1C) test, (2) the percentage of patients with at least 1 dilated retinal examination, and (3) the percentage of patients with at least 1 microalbuminuria or clinical albuminuria test.

 Humanistic metrics included patient self-assessment of change in quality of life, improved understanding of both the disease and self-management, and overall program satisfaction.

 Patient characteristics and demographics came from the Defense Enrollment Eligibility Reporting System. Healthcare utilization and expenditures came from TRICARE’s Medical Data Repository. These included care provided through TRICARE’s purchased care program and care provided at military treatment facilities. Information on DM services received and patient DM participation status came from the MCSC patient tracking system.

 A disease-specific questionnaire was mailed to patients 6 months after their initial baseline assessment to collect information on change in quality of life, perceived usefulness of the program and materials received, and overall program satisfaction. The survey questionnaire (available from the authors) is designed based on existing survey instruments such as the SF-36 short-form survey, Healthcare Effectiveness Data and Information Set Member Satisfaction Survey, and the 2006 Disease Management Association of America Participant Satisfaction Survey.

 A total of 73,156 patients were contacted by the MCSCs as of September 30, 2008. The analysis excluded 8969 patients who had opted out of the program and 4685 patients with fewer than 6 months of program tenure (who had insufficient time to complete the program). Also excluded from the analysis were patients whose complete medical records were no longer available because the patient became Medicare eligible (n = 434), was no longer eligible for TRICARE (n = 1272), or died (n = 25). Patients diagnosed with HIV/AIDS (n = 50) or end-stage renal disease (n = 232) were excluded, with the exception of diabetes patients first diagnosed with end-stage renal disease after becoming eligible for DM.

 As these are opt-out DM programs with eligibility determined by high utilization, using nonparticipants as the control group to measure program effect introduces the problem of selection bias, whereas a pre—post evaluation design introduces the problem of regression to the mean.

 To reduce these known biases, we identified a comparable sample of patients, referred to as the historical control group (HCG), and used their outcomes to predict outcomes for the DM population in the absence of a DM program.

 The HCG consists of 12,192 asthma, 3247 CHF, and 23,778 diabetes patients who in September 2004 met the same eligibility criteria used to identify candidates for DM in September 2006 and later. We estimated multivariate regressions using patient annual outcomes from October 2004 to September 2005 as the dependent variables. We used Poisson regression to model inpatient days and ambulatory visits, a generalized linear model with gamma distribution to model medical expenditures, and logistic regression to model receipt of tests and pharmaceuticals. Separate regressions were estimated for the Prime (managed care) and non-Prime (preferred provider organization) populations. Regression results are available from the authors upon request.

 Covariates included patient characteristics in September 2004 (age, sex, military sponsoring service [Army, Navy, Air Force, other], region [North, South, West]) and utilization between October 2003 and September 2004 (Charlson Comorbidity Index and disease-related inpatient days and ambulatory visits). For asthma we included the number of 30-day asthmarelated prescriptions. For CHF we included whether the patient used ACE inhibitors and beta-blockers. For diabetes we included whether the patient had an A1C test in the past year and used total rather than diabetes-specific utilization.

 Using regression coefficients estimated with the HCG, we predicted DM patient outcomes for the period following program eligibility based on patient characteristics when first eligible for DM and healthcare utilization patterns during the 12 months prior to DM. Differences between annualized observed and predicted outcomes for the DM population were attributed to DM.

DM Process and Patient Participation Outcomes

 Although the regression analysis controls for observed differences between the DM and HCG populations, the populations are relatively similar (

). The DM population tended to be slightly less healthy as measured by the Charlson Comorbidity Index, had higher rates of ambulatory visits, and had lower rates of emergency visits and inpatient days during the 12-month identification window.

METHODS

Findings from TRICARE's disease management programs for asthma, congestive heart failure, and diabetes patients suggest that the programs more than pay for themselves.

Outcomes and Lessons Learned From Evaluating TRICARE's Disease Management Programs

 To compare changes in medication use and costs over time for management of painful diabetic peripheral neuropathy (pDPN) or postherpetic neuralgia (PHN) among patients in commercial health plans requiring prior authorization (PA) for pregabalin versus patients in plans without pregabalin PA policies.

Retrospective claims data were obtained for 2005 to 2007 from 6 health plans with pregabalin PA and 6 health plans without pregabalin PA. Differences in resource utilization and costs were compared between baseline and 1-year follow-up periods using a pre—post parallel-group design.

Adults diagnosed as having pDPN or PHN with at least 1 claim for pDPN- or PHN-specific pain medication were selected. Pharmacologic therapy, healthcare utilization, and expenditures were analyzed using bivariate statistics and generalized linear models via a difference-in-differences approach comparing cohorts year over year.

The 2 cohorts included 2084 patients in PA plans and 1320 patients in non-PA plans. Compared with non-PA plans, plans requiring PA experienced a 5.0—percentage point lower increase in patients using pregabalin year over year (P <.001). Utilization in PA plans of other anticonvulsants was 3.7 percentage points higher (P = .03), while nonopioid analgesic use was 5.2 percentage points lower (P = .01). There were no significant differences in opioid, antidepressant, or other pDPN or PHN medication use or pDPNor PHN-related total healthcare costs.

A PA policy for pregabalin was associated with lower pregabalin utilization, but there was no statistically significant effect on pDPN- or PHN-specific medication or healthcare expenditures.

This study compared healthcare utilization and expenditures between health plans with versus without a prior authorization (PA) policy for pregabalin.

Plans with PA had significantly lower proportions of patients with any pregabalin use. 

Plans with PA experienced higher use of other anticonvulsants; lower use of nonopioid analgesics; no differences in the use of opioids, antidepressants, or other painful diabetic peripheral neuropathy (pDPN)- or postherpetic neuralgia (PHN)-specific drugs; and no net cost effect. 

Although the PA policy accomplished the objective of controlling access to pregabalin, there were no statistically significant differences in pDPN- or PHN-specific medication or healthcare expenditures.

	Health plans and government-sponsored healthcare programs have been under constant and increasing pressure to balance costs of care with improved health outcomes and high-quality care.

 This is particularly true for managing prescription drug expenditures, one of the most visible sectors of healthcare spending.

 Various measures intended to strike that balance, including generic substitution, tiered copayment or cost sharing, step therapy, and prior authorizations (PAs),

 have been widely adopted by health plans and pharmacy benefit managers. Prior authorization, a requirement to obtain prior approval from the health plan or pharmacy benefit manager for prescription medication reimbursement, is recognized for its potential as an effective tool for encouraging appropriate and cost-effective drug utilization, reducing unnecessary prescription drug use, and controlling expenditure growth.

 have shown mixed evidence on the overall effect of PA policies, with some documenting potential for savings and others showing negative clinical and economic effects. Given the administrative costs of running the PA programs and the burden they put on healthcare providers (physicians, nurse practitioners, and pharmacists) and patients, managed care organizations are prudent to evaluate whether their PA programs are meeting the intended objectives.

 investigated the effect of PA policies on Medicaid patients diagnosed with postherpetic neuralgia (PHN) or having neuropathic pain associated with diabetic peripheral neuropathy (pDPN), comparing states with versus without such PA policies. Neuropathic pain associated with diabetic peripheral neuropathy, a complication of diabetes mellitus,

 and PHN, a complication of herpes zoster infection,

 are 2 neuropathic pain conditions that share common pharmacologic treatments, including opioids, antidepressants, and anticonvulsants.

 Although pregabalin (Lyrica [schedule V]; Pfizer Inc, New York, NY) is one of the medications indicated for and recommended as first-line treatment in pDPN and PHN,

 many health plans have restricted access to it by requiring PA for reimbursement. The study by Margolis et al

 found that, although state Medicaid programs’ pregabalin PA requirements accomplished the objective of lower pregabalin utilization, they were associated with the unintended effects of increased opioid use and disease-related healthcare costs. Because that study was undertaken in a Medicaid population among whom patient characteristics and policies are likely to differ from those in commercial health plans, the present study was undertaken to investigate the effect of a pregabalin PA policy among commercially insured patients with pDPN and PHN.

	The primary objective of this study was to investigate the overall effect of PA policies restricting access to pregabalin on utilization of pregabalin and other pharmacologic therapies. We also evaluated the difference in overall healthcare utilization and associated direct costs for the management of pDPN or PHN between patients in commercial health plans with versus without PA policies for pregabalin.

	The intervention evaluated was the introduction of a restriction on the use of pregabalin for the treatment of pDPN and PHN in the form of PA compared with unrestricted access to pregabalin for these indications. The effect of PA was assessed in commercial health plans using a pre—post parallel-group design. Calendar year 2005 was designated as the baseline (preindex) period. A 6-month PA initiation window followed the baseline period. The 1-year postindex follow-up period began on July 1, 2006, and ended on June 30, 2007. The study sample was limited to patients with a diagnosis of DPN or PHN at any point from 2005 to 2006. Evidence of treatment for symptomatic relief of DPN or PHN pain was required as a proxy for pDPN (all PHN is acknowledged as painful). Utilization and expenditure data were compared between the baseline and follow-up periods for the 2 cohorts of health plans with versus without PA for pregabalin. The study tested the following 3 hypotheses: (1) Access restrictions to pregabalin in the form of PAs are associated with lower utilization of pregabalin relative to a comparable population without such restrictions. (2) The use of PA for restricting pregabalin use in health plans is associated with higher utilization of other pharmacologic treatments for pDPN or PHN. (3) The use of PA for restricting pregabalin use in health plans is associated with higher overall and disease-related healthcare service utilization and expenditures for the management of pDPN or PHN.

	Years 2005-2007 of the Thomson Reuters MarketScan Commercial Claims and Encounters Database were used for this study. This database includes complete longitudinal records of inpatient services, outpatient services, long-term care, and prescription drug claims covered under various fee-for-service and capitated health plans, including exclusive provider organizations, preferred provider organizations, point-of-service plans, indemnity plans, and health maintenance organizations. Retrospective data were obtained from 12 commercial health plans, 6 that had implemented PA around the time pregabalin was launched and 6 with no known restriction on reimbursement for pregabalin. Plans with step therapy, quantity limits, or formulary restrictions other than PA were not included. Plans with tiered formularies were included. Health plan restrictions (PA, step therapy, etc) were not examined for pDPN- or PHN-related medications other than pregabalin. The health plans’ identities cannot be divulged because of contractual arrangements involved in acquiring their data.

	Patients were required to be a minimum of 18 years of age and continuously enrolled in the health plan with both medical and pharmacy benefits from January 1, 2005, through June 30, 2007. They were required to have at least 1 claim with an 

International Classification of Diseases, Ninth Revision, Clinical Modification

) diagnosis code for DPN (250.6x or 357.2x) or PHN (053.1x) at any point from 2005 to 2006 and at least 1 claim within 60 days after a DPN or PHN diagnosis from 2005 to 2006 for a medication used in treating pDPN or PHN (

). Patients with pDPN or PHN were analyzed as one group because of the small PHN sample size (in both cohorts, approximately 86% of patients had only pDPN).

	Patients with Medicare coverage were excluded to represent the perspective of a commercially insured employed population. Patients were also excluded if they had resided in a long-term care facility for 90 days or longer or if they had any claims with an ICD-9-CM diagnosis code for epilepsy (345.xx or 780.39), 1 inpatient or 2 outpatient nondiagnostic claims for cancer (140.xx-172.xx, 174.xx-208.xx, or 235.xx-239.xx, except basal cell and squamous cell skin cancers and benign neoplasms), or transplant surgery (codes available on request from the corresponding author) at any point during the study period. Table 3 summarizes the incremental attrition associated with each of these criteria and the resulting cohort sample sizes for PA and non-PA health plans.

	The study analysis period began on January 1, 2005, and ended on June 30, 2007. Calendar year 2005 was used as a baseline period. A 6-month PA initiation window from January to June 2006 followed the baseline period to minimize possible misclassification of data when the exact starting dates of the PA policies were not known and in recognition of the fact that the implementation date for PA would vary across plans. The 1-year postindex follow-up period ran from July 1, 2006, through June 30, 2007. The use of 1 year of follow-up data was chosen to provide equal lengths in the preindex and postindex periods. Pregabalin was approved for treatment of pDPN and PHN in the United States in September 2005; however, the volume of claims for the few patients receiving pregabalin in 2005 (3.7% of patients) was low (1.7 prescription claims per patient receiving pregabalin).

	Diagnoses and pharmacologic therapy were derived from medical and pharmacy claims received according to service dates in the study period. Drugs for treating pDPN or PHN were analyzed by drug class as summarized in Table 1 and for pregabalin individually. The pDPN or PHN diagnoses and comorbidities were determined from nondiagnostic medical claims.

	Healthcare utilization data included hospital admissions, length of stay, emergency department visits, physician office visits, other outpatient services, and numbers of outpatient prescriptions by class and for pregabalin individually. Expenditures were adjusted to 2007 constant US dollars, measured as the gross covered payments for all healthcare services or products (ie, the amount eligible for payment after applying pricing guidelines such as fee schedules and discounts and before applying deductibles, copayments, and coordination of benefits). Utilization and expenditures were measured for all medical and pharmacy claims. Medical claims were designated as DPN or PHN specific when a diagnosis for DPN or PHN appeared in any position in the claim.

	Bivariate descriptive statistics were used to characterize the study population in terms of all key dependent, independent, and control variables by year. All dependent and independent variables were compared in the 1-year postindex period (July 1, 2006, to June 30, 2007) with the 2005 baseline within cohort, and the year-over-year differences were contrasted between cohorts as follows: Difference in Differences = (Restricted Cohort 2006-2007 — Restricted Cohort 2005) – (Unrestricted Cohort 2006-2007 – Unrestricted Cohort 2005).

 χ2 Tests were used to evaluate differences for categorical variables, and t tests were used for continuous variables. Tests for statistically significant differences in outcome variations of PA plans versus non-PA plans before and after the intervention (difference in differences) were conducted. A significance level of .05 was used to determine statistical significance.

	Multivariable analyses were used to explore the association between PA restrictions for pregabalin and the outcomes of interest. A logit model was used to compare pregabalin use in the postindex period between the plans with versus without PA restriction. The logit model used a dichotomous outcome variable indicating the use of pregabalin in the postindex period and a stochastic error term to capture nonsystematic variation. The independent variable of interest was the indicator for enrollment in a health plan with a PA policy.

	Generalized estimating equation models were used to test the hypotheses of differential increments in outcomes between plans with versus without PA after controlling for the potential confounding effects of age, sex, geographic region, health plan type, other chronic pain conditions, and comorbidities. Cluster units were defined at the individual level because a given patient contributed data to the preperiod and postperiod. The following outcomes were included in the analysis: opioid use, nonopioid analgesic use, other pDPN or PHN medication utilization, and pDPN- or PHN-related healthcare costs.

	The key independent variables in the regressions included an indicator for enrollment in a health plan with a PA policy, a time indicator (for 2006-2007 vs 2005), age, sex, geographic region, health plan type (health maintenance organization, preferred provider organization, point of service, or other), and a set of diagnostic flags to control for the effect of comorbidities. In linear regression models, the difference-indifferences estimate was the coefficient associated with the interacted term for a health plan with a PA policy and the period indicator. Because this coefficient has no direct interpretation in nonlinear models, the marginal effects of the interaction terms were calculated to express these effects across the multivariable models using the same units of measure as the original dependent variable (eg, probability of use, number of claims, and US dollars).

	Among the generalized estimating equation models, logit link and binomial variance functions were used for modeling opioid use. Log link and negative binomial variance functions were used for counts outcomes (ie, utilization of nonopioid and other pDPN or PHN medications). Log link and gamma variance functions were used for pDPN- or PHN-related costs. The variance functions for the generalized estimating equation models were verified based on the results of the modified Park test on the raw scale residuals. The exchangeable correlation structure was implemented and verified by the quasi-likelihood under the independence model criterion as the best working correlation structure. Analyses were conducted using STATA release 9 (StataCorp LP, College Station, TX).

	A total of 3404 patients met all subject selection criteria (Table 3), with 2084 patients in PA plans and 1320 patients in non-PA plans. In both cohorts, approximately 86% of patients were diagnosed as having DPN and approximately 14% as having PHN (

). The PA cohort was slightly older (mean [SD] age, 52.3 [6.9] vs 51.6 [7.2] years, P = .004), with members residing primarily in the western and north central regions of the United States, compared with the non-PA cohort, whose members were located mostly in the southern region. Insurance coverage differed between cohorts, with mostly preferred provider organization and health maintenance organization insureds in PA plans, while non-PA members had predominantly preferred provider organization coverage and point of service to a lesser extent. Sex distribution and preindex comorbidity were not significantly different between the 2 cohorts. Analysis of preindex medication use (antidepressants not used in pDPN or PHN treatment, anxiolytics, hypnotics, migraine medications, and muscle relaxants) also showed no statistically significant differences between cohorts.

POLICY

Prior authorization for pregabalin in commercial insurance plans accomplished the objective of lower pregabalin utilization; however, there was no significant decrease in disease-related healthcare costs.

Healthcare Utilization and Cost Effects of Prior Authorization for Pregabalin in Commercial Health Plans

With enactment of healthcare reform legislation now a fact, attention has quickly turned to its implementation and necessary preparations for the changes to come. Many of the provisions of the new law will not take effect for several years, but others have more aggressive time lines and no doubt are already receiving considerable attention within federal agencies that will need to prepare for those changes. One of these changes is the creation of a nonprofit comparative effectiveness research (CER) entity known as the Patient-Centered Outcomes Research Institute (PCORI). Although it will not be an agency within the federal government, PCORI will benefit from the significant CER effort currently under way as a result of the 2009 American Recovery and Reinvestment Act.

 The American Recovery and Reinvestment Act not only expanded the federal commitment to CER, but also required specific initiatives to define CER and to develop recommendations on priority areas for new research. These important activities will further inform the process and functions of the new PCORI.

Need for Credible Real-World CER Evidence

 The American Recovery and Reinvestment Act called for a study by the Institute of Medicine to define priorities for CER

 and also established the Federal Coordinating Council for Comparative Effectiveness Research, which was charged with coordination of all federal CER activities. Although this council was terminated under the new healthcare reform law, it did establish a definition for CER that is widely accepted and expected to be embodied in the work of PCORI. That definition includes the concept that CER should provide evidence on real-world effects of treatments, processes, and technologies—that is, the effects of medical care in the usual clinical setting.

 Although prescription drugs are subject to stringent requirements for regulatory approval and can be marketed only after demonstrating safety and efficacy in prospective, well-controlled trials, regulatory approval does not require demonstration of the effects in usual care settings. In addition, for some medical procedures, ethical and practical considerations may limit the opportunity for assessment of comparative effectiveness in formal clinical trials. Therefore, the challenge of providing information on the real-world effects of treatments and procedures is significant and should be addressed comprehensively with attention to rigorous scientific standards.

 Many consider the prospective randomized controlled trial (RCT) to be the gold standard for clinical evidence. However, challenges associated with the RCT as the principal method for CER include the cost and time required for conducting such studies and their potential for broad generalizability. These challenges make RCTs a problematic standard for the range and types of real-world evidence that are needed in a new era of CER. Increasingly, attention is turning to nonexperimental study designs using observational data sets that in some cases are derived from the records compiled from the daily interactions of patients and providers in the typical clinical office setting. In addition to data generated from electronic health records, which will increase in importance as a result of new government incentives for adoption of such systems, these data sets more commonly include the data from claims processing or disease registries. How best to use these data to inform the CER enterprise is a question of critical importance that is, as yet, unanswered.

 Research using observational data already occurs frequently, and has for a long time. Using such data for research is clearly consistent with the Institute of Medicine’s evolving concept of a “learning healthcare system” in which healthcare delivery continuously benefits as real-world evidence accumulates.

 Observational data are plentiful and readily available, and a growing number of payer organizations are recognizing the valuable resource available to them for understanding how treatments work in practice. But the concept of “good,” “sound,” or even “credible” observational research is not well understood or agreed upon, and this lack of agreement can result in uneven and unpredictable consideration of such research when making coverage and payment decisions. Lack of predictability wastes resources both for those reviewing and interpreting the results of observational research and for the suppliers of that research.

Reaching Stakeholder Consensus on Standards for Good Research

 Payers and researchers alike should strive for a consistent view of what constitutes good observational research, especially when it assesses the comparative benefits of medical treatment options. With the ever-increasing demand for answers to questions about the comparative benefits of medical treatments, the tasks of researchers who generate comparative evidence and payers who interpret it should be efficient and timely. Efforts in recent years have focused on standardizing the format in which evidence is presented for consideration by payers,

 and these efforts have enhanced the efficiency of reviewing and considering evidence. But broadly accepted criteria and standards for the quality of research evidence can improve both the efficiency of the process and the quality of patient care.

 The new healthcare reform legislation will establish a methodology committee that is likely to address many of these issues, but this committee will be created only after PCORI is established. The methodology committee then has 18 months to develop its recommendations for methodologic standards. The work of this committee can be aided if stakeholders come together to review the experience with observational research to date, share successes and failures with this approach, identify the methodologic gaps that limit the utility of observational comparative research, and agree on at least the outline of appropriate standards for CER using observational data. A cooperative effort toward this objective can contribute meaningfully to the work of the new methodology committee and in the near term can enhance the predictability of the quality of the research that is offered in support of medical treatments and therapies. Such an effort also can improve the predictability of the interpretation of this research by those who must make decisions about coverage and payment.

 An initial effort intended to improve the quality and usefulness of this type of research is described in this issue of 

 My organization, the National Pharmaceutical Council, provided the initial funding for the independent development of the principles for Good Research for Comparative Effectiveness (GRACE), which addresses many of the quality issues facing CER when observational data are used. These principles are not prescriptive and do not address or suggest solutions to unsolved methodologic problems, but they do focus on the issues that should be addressed or at least discussed when observational CER is reported or presented. They leave to the reviewer the task of judging the quality and thoroughness of the response to these issues in any given piece of research. By identifying the key issues to be addressed, the principles suggest a framework for addressing the quality of observational CER and may serve as the starting point for a broad-based, multistakeholder focus on this type of research.

 Demand continues for more and better evidence about the practical and real-world effects of medical treatments. Data sources that capture information about the healthcare experience continue to proliferate. As a result, there is an urgent need for consistency and harmonization of standards for evidence generation based on these data, as well as for interpretation and application of the findings that result from the research. Patients will be the ultimate beneficiaries of a successful effort to harmonize the generation, interpretation, and application of real-world evidence.

Gary Persinger and Les Paul, MD, MS, of the National Pharmaceutical Council contributed to this commentary.

From the National Pharmaceutical Council, Washington, DC.

The National Pharmaceutical Council provided funding to Outcome, Inc in support of the GRACE principles (see the article by Dreyer et al, in this issue [p 467]).

Mr Leonard reports serving as a board member for the Health Industry Forum, the National Health Council, and the Pharmacy & Therapeutics Society.

Concept and design; drafting of the manuscript; critical revision of the manuscript for important intellectual content; administrative, technical, or logistic support; and supervision.

Daniel T. Leonard, MA, President, National Pharmaceutical Council, 1501 M St, NW, Washington, DC 20005. E-mail: 

Rigorous scientific standards are needed to address the challenge of providing information on the real-world effects of treatments and procedures.

June 2010

Prevalence of Osteoporosis in Men in a VA Rehabilitation Center

Is There a Relationship Between Early Statin Compliance and a Reduction in Healthcare Utilization?

Outcomes and Lessons Learned From Evaluating TRICARE's Disease Management Programs

Healthcare Utilization and Cost Effects of Prior Authorization for Pregabalin in Commercial Health Plans

Comparative Effectiveness and Real-World Evidence

The Shaky Foundation of the Patient-Centered Medical Home

Using Simulation to Identify and Resolve Threats to Patient Safety

Ambulatory Care Visits for Pap Tests, Abnormal Pap Test Results, and Cervical Cancer Procedures in the United States

GRACE Principles: Recognizing High-Quality Observational Studies of Comparative Effectiveness

Care Fragmentation and Emergency Department Use Among Complex Patients With Diabetes