Impact of Initial Days' Supply of Angiotensin II Receptor Blockers

, , , ,
The American Journal of Pharmacy Benefits, January/February 2016, Volume 8, Issue 1

For ARB-class antihypertensive agents, we examined the association among initial prescription supply days (ie, 30- or 90-day) and adherence, hospitalization risk, and healthcare use.

ABSTRACT

Objectives: To compare adherence, risk of hospitalization, and healthcare resource use among patients initiated with a 30- versus 90-day supply of angiotensin II receptor blockers (ARBs).

Study Design: Retrospective cohort study using administrative claims data.

Methods: Hypertensive patients 18 years or older, who initiated 30- or 90-day ARB prescriptions between January 1, 2006, and April 1, 2011, were identified and followed for 1 year. We measured adherence (proportion of days covered [PDC]; suboptimal adherence = PDC <80%), discontinuation, and healthcare resource utilization and costs. Survival analysis evaluated associations between initial days’ supply and hospitalization, controlling for age, gender, and previous hospitalization.

Results: There were 487,380 subjects included in the study. The 90-day cohort (n = 361,943) was older, had more comorbidities, and fewer previous hospitalizations. Patients initiated on 90-day prescriptions had a higher mean 1-year PDC (79.9% [95% CI, 79.8%-80.0%] vs 63.7% [95% CI, 63.7%-63.8%]); less frequent suboptimal adherence (53.2% [95% CI, 53.1%-53.4%] vs 70.6% [95% CI, 70.5%-70.6%]); and a lower 1-year discontinuation rate (32.6% [95% CI, 32.4%-32.8%] vs 45.5% [95% CI, 45.4%-45.6%]). Inpatient and emergency department (ED) costs were $2301 and $314 less, respectively, for inpatient and ED visits among patients without prior hospitalization initiated on a 90-day supply.

Conclusions: Patients initiated on a 90-day ARB supply showed improved adherence, lower discontinuation rates, lower hospitalization risk, and lower inpatient/ED costs compared with patients initiated on a 30-day supply.

Am J Pharm Benefits. 2015;8(1):e1-e8

PRACTICAL IMPLICATIONS

Our model estimated the total annual stroke-related healthcare costs for a health plan to be $4,930,787, with 74% of costs attributable to ischemic stroke (IS) due to its higher incidence and lower mortality rate.

  • Despite the elevated risk of stroke among patients with atrial fibrillation, few studies have been published that assess the relative burden of IS to hemorrhagic stroke among this population.
  • Our model can be used by health plans to estimate the costs of IS and hemorrhagic stroke among atrial fibrillation patients.
  • These data provide additional insights into future risk-benefit assessment of oral anticoagulation therapy from an economic perspective.

An estimated 77.9 million Americans have hypertension,1 putting them at risk for heart disease and stroke, which are leading causes of morbidity and mortality in the United States.2,3 Costs attributable to hypertension total nearly $131 billion per year for medical care and $25 billion for lost work productivity.4 While 74.9% of US adults with hypertension are prescribed antihypertensive medications, only half (52.5%) of those individuals achieve clinical control.1

A leading cause of suboptimal blood pressure control is nonadherence to antihypertensive medications.2,5-7 Estimates of suboptimal adherence to antihypertensive drugs during the first year of therapy range from 30% to 69%.5,6,8-10 Suboptimal adherence with antihypertensive medications is associated with increased risk of stroke, cardiovascular disease, myocardial infarction, and increased healthcare utilization and medical costs.2,5,6,7

Medication nonadherence is a multifaceted problem that may be related to cognitive issues, patient—doctor relationships, patient decision making, attitudes toward medications, cultural issues, socioeconomic factors, health-system related factors, and medication characteristics.5,6,11 Research has consistently demonstrated that patient adherence improves as the burden of taking the medication is reduced. For example, medication adherence increases when a thrice-daily dose is reduced to a once-daily dose.5,12 The burden of taking medications encompasses far more complexity than high dosing frequency, yet little is known regarding other ways to reduce the burden, such as reducing trips to the pharmacy by increasing the days’ supply of each prescription.

While many potential causes of nonadherence have been studied,5,6,11 the impact of prescribed drug quantity on medication adherence has only recently been investigated. A California-based medical claims study found 20% higher adherence associated with 90- versus 30-day supply of prescriptions among those prescribed antihypertensive medications, statins, selective serotonin reuptake inhibitors, or oral hypoglycemic medications.13

Similarly, an Illinois claims-based study found that patients with a 90-day supply of hypertension, diabetes, or hypercholesterolemia medication were 7% to 10% more likely to adhere to treatment than those prescribed a 30-day supply.14 While these studies suggest an association between higher initial prescription supply and improved adherence, the potential impact on healthcare resource utilization has not been previously investigated.

The purpose of this study was to examine the effect of the initial prescription of angiotensin II receptor blocker (ARB) therapy for a supply of 30 or 90 days on adherence, discontinuation, risk of hospitalization, healthcare resource utilization, and associated costs among a large, geographically diverse hypertensive patient population.

METHODS

Data Sources

Data from Truven Health Analytics MarketScan Commercial Claims and Encounters and Medicare Supplemental Databases were used for the time period of January 1, 2005, through March 31, 2012. The MarketScan database has been validated and widely used for research purposes, and contains over 500 million medical and pharmacy claims per year from individuals with private healthcare insurance. The data represent approximately 45 large employers who self-insure their employees and dependents.

Design

The study was a retrospective cohort analysis of hypertensive patients using administrative claims data.

Sample Selection

The patient selection flowchart is shown in Figure. Study patients were 18 years or older with a diagnosis of essential hypertension [International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes 401.xx], no ARB use during a 12-month baseline period, continuous enrollment during the 12-month baseline period and during the 12-month follow-up period after the date of first ARB prescription (designated as the index prescription). Patients were categorized into 2 study cohorts by their index ARB prescription: those who received 30-day supplies and those who received 90-day supplies.

Baseline Comparisons

We compared the cohorts with respect to the following baseline characteristics: age, gender, region of residence, insurance plan type, hospitalization history (any admission), and comorbidity burden (calculated using the Diagnostic Cost Group Hierarchical Condition Category [DxCG] classification system).15 The DxCG risk adjuster creates a single score for each individual based on the diagnosis fields of claims records. The risk score was then normalized, such that a score greater than 1 indicates a population that is sicker than average and a score lower than 1 indicates a population that is healthier than average.

Outcome Measures

Outcomes of interest were measured during the first year of follow-up (ie, 365 days after index prescription date) which included the following: adherence, defined as the proportion of days covered (PDC: the number of days’ supply dispensed, divided by the number of days in the observation period); proportion with suboptimal adherence (defined as PDC <80%); rate of discontinuation (≥60 consecutive days without a refill); number of inpatient admissions, emergency department (ED) visits, and outpatient visits (all-cause and hypertension-specific, defined as ICD-9-CM code 401.xx as diagnosis for the medical claim); and costs associated with healthcare resource utilization (including prescription medications).

Statistical Analyses

Descriptive analyses were conducted to compare baseline demographics by days’ supply of the index ARB. Proportions, group means, and 95% CIs were calculated, with differences between groups tested for statistical significance using

χ

2 tests and t tests. We used survival analysis models16 to estimate adjusted hazard ratios (HRs) for time to discontinuation and time to hospitalization.

The model included time-fixed terms for the index days’ supply indicator, sex, age, and age-squared. The quadratic term was included to account for nonlinearity identified in diagnostic analyses. We estimated separate models for populations with and without previous hospitalizations, since we hypothesized that patients with previous hospitalizations may be more likely to be readmitted and may skew estimates related to healthcare utilization.

Both all-cause and hypertension-related costs associated with healthcare resource utilization (ie, costs for inpatient, ED, and outpatient visits) as well as all prescription drugs (both ARBs and non-ARBs) were estimated. Mean costs were estimated by healthcare utilization type (ie, inpatient, outpatient, ED, and prescription drug use) among those patients who utilized any services within that category during the study period. We used 2-point finite mixture models to estimate the adjusted differences in spending by initial days’ supply.17,18 All costs were inflated to 2012 using the Consumer Price Index.

Multivariate models were tested for the proportional-hazard (PH) assumption. Due to the establishment of a stringent criterion of meeting the PH assumption, a parsimonious approach towards choosing covariates was adopted. Statistical significance was defined at the significance level of P <.05. Several sensitivity and sub-analyses were also conducted, including stratifying the proportional hazards models by census region and insurance type due to differences at baseline. All analyses were conducted using Stata version 10 (StataCorp LP, College Station, Texas) and SAS version 13.2 (SAS Institute, Cary, North Carolina).

This study utilized de-identified patient information and thus was exempt from institutional review board approval requirements.

RESULTS

Sample Size

A total of 487,380 subjects met the selection criteria, of whom 361,943 (74.2%) received an initial 30-day supply of drug, and 125,437 (25.8%) an initial 90-day supply (eAppendix, available at www.ajmc.com). Within the 90 days’ supply study group, only 11% of patients had more than 1 prescription with a different days’ supply during the study follow-up period.

Demographics

The 90-day cohort, compared with the 30-day cohort: 1) was older, with a mean age of 62.0 vs 55.4 years; 2) differed by region of residence (31.7% vs 22.7% from the midwest, and 42.5% vs 51.6% from the south); 3) had different insurance coverage (26.9% vs 12.4% with comprehensive, and 8.7% vs 15.7% in a health maintenance organization [HMO]); 4) were less likely to have experienced a hospitalization in the previous year (13.5% vs 15.2%); and 5) had a higher mean comorbidity risk score (1.10 vs 0.97). All differences were statistically significant at the P <.05 level (Table 1).

Adherence

Adherence was consistently higher in the 90-day cohort than the 30-day group at 6, 9, and 12 months (Table 2). Suboptimal adherence (cumulative PDC <80%) was significantly lower for the 90-day cohort than for the 30-day cohort at 6, 9, and 12 months (32.9% vs 58.8%, 45.5% vs 66.0%, and 53.2% vs 70.6%, respectively; P <.001). Cumulative discontinuation rates were lower with the 90-day cohort at 6, 9, and 12 months (Table 2). All comparisons were statistically significant at the P <.001 level.

Risk of Drug Discontinuation During the 1-Year Follow-up

Patients initiated on a 90-day supply of ARBs versus a 30-day supply had a significantly lower 1-year discontinuation rate (32.6% [95% CI, 32.4%-32.8%] vs 45.5% [95% CI, 45.4%-45.6%]).

Risk of Hospitalization During the 1-Year Follow-up

The risk of hospitalization was lower among patients initiated on a 90- versus a 30-day supply for patients with and without prior hospitalization (adjusted HR, 0.89 [95% CI, 0.86, 0.93; P <.001]; and adjusted HR, 0.97 [95% CI, 0.95-1.00; P = .031 for with and without prior hospitalization, respectively) (Table 3). Sub-analysis by geographic region and by health insurance plan type found all estimates of the main model remained unchanged as well as statistically significant, with the exceptions of: cohort comparison in the West region (adjusted HR, 0.95; 95% CI, 0.89-1.01); and the comparison in the subgroup of persons with point-of-service (POS) insurance (adjusted HR, 0.98; 95% CI, 0.91-1.06).

Healthcare Resource Use

For “hypertension-related medical services,” resource utilization rates per 100 patients were lower for the 90-day cohort versus the 30-day cohort for inpatient visits (3.8 vs 4.4; P = .019), ED visits (4.6 vs 6.7; P = .011), and outpatient visits (76.2 vs 81.5; P <.001). For “all-cause medical services,” resource utilization rates were significantly lower for the 90-day cohort than the 30-day cohort for ED visits (21.4 vs 23.8; P = .008), but higher for outpatient visits (98.7 vs 98.5, P = .04).

Reduced all-cause inpatient and ED utilization associated with 90-day initial ARB prescription was observed among patients with prior hospitalizations (Table 4). Patients with prior hospitalization and 90-day initial ARB prescriptions had lower hypertension-related utilization across all categories (Table 4). During the first year, among patients without prior hospitalizations, all-cause inpatient visit rates (per 100 patients) were higher (10.8 [95% CI, 10.58-10.95] vs 9.3 [95% CI, 9.16-9.36]; P <.001) and ED visit rates were lower (19.0 [95% CI, 18.78-19.25] vs 20.4 [95% CI, 20.30-20.59]; P <.001) for patients with an initial 90-day supply compared with 30-day supply; outpatient visit rates differed by a small, though statistically significant, magnitude (—0.3) (Table 4).

Among patients without prior hospitalizations, hypertension-related ED visit rates (per 100 patients) (4.2 [95% CI, 4.06-4.30] vs 6.0 [95% CI, 5.89-6.06]; P <.001) and outpatient visit rates (76.4 [95% CI, 76.19-76.69] vs 81.8 [95% CI, 81.70-81.97]; P <.001) were lower among patients with 90-day initial supply, while inpatient visit rates were similar.

Costs

Among patients with prior hospitalizations, all-cause inpatient costs were lower among patients initiating ARB therapy with a 90-day versus 30-day prescription (by $7071), while hypertension-related inpatient costs were higher (by $770) (Figure). Among patients without prior hospitalization, some costs associated with all-cause medical services were significantly lower for the 90-day cohort versus the 30-day cohort, including inpatient visits ($2301 lower per patient) and ED visits ($314 lower per patient), but outpatient visit costs were $157 higher (Figure).

For the same subpopulation (Figure), all the hypertension-related medical services costs were lower for the 90-day cohort compared with the 30-day cohort, including $216 lower for inpatient visits, $196 lower for ED visits and $92 lower for outpatient visits costs per patient during the first year (all statistically significant at the P <.05 level). In general, the prescription drug costs were higher for the 90-day supply group than the 30-day supply group.

DISCUSSION

Medication adherence is of great concern to clinicians, healthcare systems, and payers because mounting evidence shows that nonadherence is prevalent and associated with adverse outcomes and higher costs of care.2,5,6,11 Our study found that initiating patients on a 90-day supply of ARBs was associated with higher adherence during the first year of therapy across a range of measures: higher refill rates, greater proportion of days covered, lower rates of suboptimal adherence, and lower rates of discontinuation, compared with initiating a 30-day supply. These improvements remained even after adjustments for age, gender, and history of hospitalization. We also found better outcomes associated with patients initiated on a 90-day supply of ARBs relative to a 30-day supply including lower risk of hospitalization in the first year and fewer visits to the ED.

These findings are consistent with a large body of evidence showing that reducing the burden of pill-taking improves adherence.5,12 Our study also contributes to early research on the relationship between the days’ supply prescription and adherence. A retrospective observational study of medical claims in California showed a 20% higher adherence associated with patients prescribed a 90-day supply of drug rather than a 30-day supply,13 and a study of claims data by BlueCross BlueShield of Illinois found patients with a 90-day supply of medication were 7% to 10% more likely to adhere to treatment after 540 days of follow-up than those prescribed a 30-day supply.14

The former study, conducted by Taitel et al,13 utilized pharmacy claims dated January 2010 from Walgreens and was conducted in a group of California Medicaid patients— 97% of whom had a 0 co-pay, which limits the generalizability of the findings. While these studies have similar findings, the populations studied were limited and also differed with respect to geographic location (California vs Illinois) and types of coverage and benefit design (Medicaid vs private insurance), and as such, each study may have limited generalizability beyond the study population.

The higher adherence associated with 90-day supply in our study is equivalent to 2 months of additional adherence per annum [(79.9% − 63.7%)

×

360 = 59.32 days]. For a hypothetical plan size of 10 million, the reduced rates (expressed as a rate per 100 patients) of hypertension-related medical services associated with 90-day supply (vs 30-day supply) as observed in our study are equivalent to about 60,000 fewer inpatient visits, 210,000 fewer ED visits, and 530,000 fewer outpatient visits. Regarding risk for hospitalization, in patients with prior hospitalization, for every 100 hospitalizations in the 30-day group, the 90-day group would incur 89 hospitalizations (P <.001).

In patients with no prior hospitalization, for every 100 hospitalizations in the 30-day group, the 90-day group would incur 97 hospitalizations (P = .031). This reduced utilization was associated with lower inpatient costs among high-risk patients (those with prior hospitalizations) in our study, as those patients initiated on a 90-day versus 30-day ARB prescription experienced an average of $7071 fewer all-cause inpatient costs. A 90-day supply may not be appropriate for all patients, as physicians may want to initially ensure adequacy of a drug for a patient.

If the drug is found to be inappropriate for a patient, a 90-day supply could potentially lead to more wastage. However, substantial benefits may be achieved by greater prescribing of 90-day supply of ARBs to those patients for whom it represents suitable therapy. Prescribing more 90-day prescriptions to appropriate patients is one way to improve adherence to antihypertensive therapy, which may lead to better clinical outcomes and ultimately lower healthcare resource utilization.

Limitations

Although we attempted to control for baseline differences between study groups, residual bias cannot be ruled out due to the observational nature of the study design. Patients were required to have a diagnosis of essential hypertension (using ICD-9-CM codes). However, since the study was conducted using claims data, it is impossible to ascertain whether ARBs were prescribed for blood pressure control or other cardiac comorbid conditions. We did not have information about the drivers of initial days’ supply of ARB prescriptions.

An important factor may be pharmacy benefit design as some prescription plans may place restrictions on the maximum days’ supply, and patients may incur different co-pay amounts. Patients initiated on 90-day supplies were older and had a higher baseline comorbidity burden, and although statistical adjustment was performed for comparisons of outcome variables, unobserved variables related to clinical risk or specific comorbidities may have influenced whether a provider prescribed 30 or 90 days’ supply.

Health plans that promote or require 90-day supply prescriptions for chronic medications may also be more likely to offer automated medication refills and this may have affected our study results. It is also possible that the 90 days’ supply patient group was more frequently prescribed generic ARBs, and thus lower drug acquisition cost could impact our medication adherence results.

Another challenge for this study was the unequal opportunity to measure the outcomes; on day 91, adherence could be measured among patients initiated on a 90-day supply prescription, compared with day 31 in the comparator group. To address this potential source of bias, we conducted a sensitivity analysis by fixing the first prescription as a 90-day supply for all study participants. Even with this adjustment the higher adherence rates held for the actual 90-day supply group (average annual PDC: 79.9% vs 70.3%; P = .002).

Also, we were unable to ascertain whether patients actually took the medication, as the claims information only provides evidence that they acquired the medication. Furthermore, this study evaluated adherence specific to ARB therapy, and did not assess whether patients may have discontinued therapy with ARBs to switch to another class of antihypertensive medication.

Finally, we did not evaluate dosage changes or treatment intensification and the potential impact of such changes on medication adherence; we also did not have detailed information regarding patient medication consumption to ascertain early medication discontinuation as opposed to nonadherence to daily prescribed medication dosing regimen (which may co-occur with continued medication persistence).

Associations between mail-order pharmacy and higher adherence have been noted in the literature.19,20 We conducted an ad hoc analysis and found differences in mail-order pharmacy use in our groups: 10.2% in the 30-day group versus 73.9% in the 90-day cohort (P <.001). To the extent our findings are due to mail-order supply, our results may be overstated.

The generalizability of our study is also limited to insured patients who face fewer cost barriers to medications. The strength of this study lies in the longitudinal design and the use of incidence drug users so that initial therapy exposure is uniformly observed.

CONCLUSIONS

In this large retrospective cohort study of patients prescribed a 30-day versus 90-day supply of an ARB, we found a 1- to 2-month adherence advantage for the 90-day supply cohort. We also found lower suboptimal adherence, lower discontinuation risk, decreased risk of hospitalization, and reduced hypertension-related medical resource utilization in the 90-day cohort. These findings suggest an initial prescription of 90-day supply of ARBs may be an effective approach for health plans to help patients improve adherence.

Acknowledgments

The authors thank Elizabeth Szamreta, MPH, fellow, Health Economics & Outcomes Research Department, Daiichi Sankyo, Inc, Parsippany, NJ, for research support.

Author Affiliations: Northeastern University (BAB), Boston, MA; University of Massachusetts Medical School (HF), Worcester, MA; Health Economics & Outcomes Research Department, Daiichi Sankyo, Inc (SK, CQ), Parsippany, NJ; University of the Sciences in Philadelphia (PPG), PA; Philadelphia College of Pharmacy (PPG), PA.

Funding Source: Daiichi Sankyo, Inc.

Author Disclosures: Dr Qian is an employee of Daiichi Sankyo, Inc. Dr Kaila is a former employee of Daiichi Sankyo, Inc. Mr Fouayzi received payment for the analysis of data and statistical analyses of this study. Dr Briesacher received a consulting fee from Daiichi Sankyo, Inc for work on this study. Dr Gerbino reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (BAB, CQ); acquisition of data (BAB); analysis and interpretation of data (BAB, CQ, HF, PPG, SK); drafting and critical revision of the manuscript for important intellectual content (BAB, CQ, HF); statistical analysis (HF); obtaining funding (BAB); administrative, technical, or logistic support (BAB); and supervision (BAB).

Send correspondence to: Becky A. Briesacher, PhD, Northeastern University, 360 Huntington Ave, Boston, MA 02115. E-mail: b.briesacher@neu.edu.

REFERENCES

1. Go AS, Mozaffarian D, Roger VL, et al; American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics—2013 update: a report from the American Heart Association. Circulation. 2013;127(1):e6-e245. doi:10.1161/CIR.0b013e31828124ad.

2. Chobanian A, Bakris G, Black H. The Seventh Report of the Joint National Committee on prevention, detection, evaluation, and treatment of high blood pressure: the JNC 7 report. JAMA. 2003;289(19):2560-2572.

3. Kochanek KD, Xu J, Murphy SL, Miniño AM, Kung HC. Deaths: final data for 2009. Natl Vital Stat Rep. 2011;60(3):1-116.

4. Heidenreich PA, Trogdon JG, Khavjou OA, et al; American Heart Association Advocacy Coordinating Committee; Stroke Council; Council on Cardiovascular Radiology and Intervention; Council on Clinical Cardiology; Council on Epidemiology and Prevention; Council on Arteriosclerosis; Thrombosis and Vascular Biology; Council on Cardiopulmonary; Critical Care; Perioperative and Resuscitation; Council on Cardiovascular Nursing; Council on the Kidney in Cardiovascular Disease; Council on Cardiovascular Surgery and Anesthesia, and Interdisciplinary Council on Quality of Care and Outcomes Research. Forecasting the future of cardiovascular disease in the United States: a policy statement from the American Heart Association. Circulation. 2011;123(8):933-944. doi:10.1161/CIR.0b013e31820a55f5.

5. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med. 2005;353(5):487-497.

6. Ho PM, Bryson CL, Rumsfeld JS. Medication adherence: its importance in cardiovascular outcomes. Circulation. 2009;119(23):3028-3035.

7. Bramley TJ, Gerbino PP, Nightengale BS, Frech-Tamas F. Relationship of blood pressure control to adherence with antihypertensive monotherapy in 13 managed care organizations. J Manag Care Pharm. 2006;12(3):239-245.

8. Vrijens B, Vincze G, Kristanto P, Urquhart J, Burnier M. Adherence to prescribed antihypertensive drug treatments: longitudinal study of electronically compiled dosing histories. BMJ. 2008;336(7653):1114-1117. doi:10.1136/bmj.39553.670231.25.

9. Nichol MB, Knight TK, Priest JL, Wu J, Cantrell CR. Nonadherence to clinical practice guidelines and medications for multiple chronic conditions in a California Medicaid population. J Am Pharm Assoc (2003). 2010;50(4):496-507. doi:10.1331/JAPhA.2010.09123.

10. Briesacher BA, Andrade SE, Fouayzi H, Chan KA. Comparison of drug adherence rates among patients with seven different medical conditions. Pharmacotherapy. 2008;28(4):437-443. doi:10.1592/phco.28.4.437.

11. Martin LR, Williams SL, Haskard KB, Dimatteo MR. The challenge of patient adherence. Ther Clin Risk Manag. 2005;1(3):189-199.

12. Claxton AJ, Cramer J, Pierce C. A systematic review of the associations between dose regimens and medication compliance. Clin Ther. 2001;23(8):1296-1310.

13. Taitel M, Fensterheim L, Kirkham H, Sekula R, Duncan I. Medication days’ supply, adherence, wastage, and cost among chronic patients in Medicaid. Medicare & Medicaid and Research Review. 2012;2(3):e1-e13. doi:10.5600/mmrr.002.03.a04.

14. Hermes M, Gleason PP, Starner CD. Adherence to chronic medication therapy associated with 90-day supplies compared with 30-day supplies [abstract]. J Manag Care Pharm. 2010;16:141-142.

15. Ash AS, Ellis RP, Pope GC, et al. Using diagnoses to describe populations and predict costs. Health Care Financ Rev. 2000;21(3):7-28.

16. Kleinbaum DG, Klein M. Survival Analysis: A Self-Learning Text. 2nd ed. New York, NY: Springer; 2005.

17. Deb P, Holmes AM. Estimates of use and costs of behavioural health care: a comparison of standard and finite mixture models. Health Econ. 2000;9(6):475-489.

18. Deb P, Trivedi PK. Demand for medical care by the elderly: a finite mixture approach. J Appl Economet.1997;12(3):313-336.

19. Adams AS, Uratsu C, Dyer W, et al. Health system factors and antihypertensive adherence in a racially and ethnically diverse cohort of new users. JAMA Intern Med. 2013;173(1):54-61. doi:10.1001/2013.jamainternmed.955.

20. Duru OK, Schmittdiel JA, Dyer WT, et al. Mail-order pharmacy use and adherence to diabetes-related medications. Am J Manag Care. 2010;16(1):33-40