Health policy experts have long touted the benefits of provider-owned insurers like Kaiser Permanente. As insurers, these provider organizations face incentives to control costs. Physicians are salaried, so they do not have incentives to recommend costly tests and procedures when less costly alternatives are available. Integration makes it easier to share information about patients and track quality, which improves the ability of the organization to manage patients with chronic conditions.

The spread of reimbursement schemes that shift risk from insurers to providers has many health system executives wondering at what point it makes sense for their systems to start selling insurance directly to consumers. However, the recent exit of several integrated plans and an antitrust ruling against 2 Chicago-area health systems that planned to start an insurer should give executives pause.

The Current Landscape of Provider-Owned Plans

Currently, there are over 100 provider-owned plans across the United States. They cover more than 26 million enrollees,

 or about 8% of the population. Kaiser Permanente is the most well-known, but there are many other regional plans, including Sanford Health Plan in the Dakotas and Optima Health in Virginia. 

Provider-owned insurers vary in terms of their relationships with outside providers and patients. Kaiser physicians provide care exclusively to the insurer’s enrollees, and those enrollees receive care exclusively from affiliated providers. Most provider-owned insurers have open networks; for example, Sanford Health Plan’s provider network includes some non-Sanford physicians and hospitals. Sanford physicians and hospitals treat patients insured by external payers in addition to plan enrollees. 

A few commercial plans have attempted to reverse-engineer the provider-owned insurer model by opening clinics. Molina Healthcare operates primary care practices in 5 states; Oscar Health announced in June 2017 that it would partner with Cleveland Clinic to sell plans in 5 Ohio counties; and United Healthcare launched a subsidiary, Harken Health, that operates primary care clinics in 6 states. In the case of Harken, although the plan included features designed to attract younger enrollees, it posted disappointing enrollment numbers, so United will close that subsidiary in 2018.

Providers considering whether to start an insurance arm face a complex decision. A new provider-owned plan will probably attract many enrollees who use the system’s hospitals and physicians and were previously insured by a commercial plan. Some of the revenue earned by the provider’s new insurance arm will be offset by decreases in revenue from traditional insurers. Whether this shift results in a net increase in profits depends on how profit margins in the insurance market compare with margins in the market for provider services.

A provider-owned health plan that disproportionally enrolls existing patients will suffer from adverse selection. Once they join a provider-owned insurer, high-cost patients shift from being revenue generators to liabilities. A successful plan needs to attract enrollees from outside the system. New provider-owned insurers lack the capacity to process claims, necessitating the hiring of a traditional insurer to pay claims to unaffiliated providers. Due to their small scale and unique structure, provider-owned insurers may have to pay more for reinsurance to protect against large losses. These costs may erode the savings from eliminating insurers as middlemen between providers and patients. Health systems that become insurers also risk creating a more antagonistic relationship with traditional insurers, who are now competitors. For example, Blue Cross Blue Shield of Nebraska, the state’s largest insurer, excluded Catholic Health’s hospitals from its provider network for 10 months after Catholic proposed to start selling insurance in the state. Catholic Health exited the insurance market in early 2017.

Employers prefer to contract with insurers that have provider networks that are accessible to all of their employees. In major urban areas, insurers need networks than span thousands of square miles. The Indianapolis metropolitan area is 6029 square miles and Phoenix’s is 9071. Provider owned insurers with only a handful of clinic locations will have trouble competing in the group insurance market. Provider-owned insurers can expand their geographic reach by contracting with unaffiliated providers, but then they forgo the benefits of integration. Smaller provider-owned insurers will be at a disadvantage relative to large commercial insurers when negotiating reimbursement rates with outside physicians and hospitals. Mergers are another option for expanding network capacity, but these are subject to antitrust review. A federal judge recently blocked a proposed merger between Chicago’s Advocate and NorthShore health systems, which planned to create a health plan. They argued that only a combined system had the geographic reach to offer a competitive product. The judge rejected as speculative the systems’ claim that the plan would save area consumers over $200 million annually.

Success Will Hinge on Ability to Cut Costs

Although some consumers appreciate the 1-stop shopping convenience of provider-owned plans, many are reluctant to enroll in plans that restrict their choice of provider. The ability of provider-owned insurers to attract enrollees and obtain favorable treatment from regulators hinges on their ability to cut costs. The RAND Health Insurance Experiment found that participants randomized to a provider-owned plan, Group Health Cooperative of Puget Sound, incurred significantly lower costs than those enrolled in fee-for-service plans.

Provider-owned plans charge higher premiums in the Medicare Advantage market,

 but our analysis of 2016 premiums in the individual market exchanges indicates that plans offered by provider-owned insurers are more affordable. The exchanges include a number of features, including market-level risk adjustment and standardization of benefits, that facilitate comparisons of premiums across plans. The 

 presents insurers’ lowest cost silver plan for a single 50-year-old consumer in the 221 exchange markets in which at least 1 provider-owned insurer sells a plan. Seventy percent of provider-owned plans’ premiums are equal to or less than the market-level median compared with only 48% of traditional plans’ premiums. 

Despite having features commonly associated with better quality and lower costs, provider-owned insurers have always operated in the shadows of the larger commercial and Blue Cross Blue Shield plans. (To paraphrase a joke about soccer in the United States, provider-owned insurers are the delivery system of the future—and always will be.) As a result of provider consolidation, there are more systems that have the scope and size to sell a closed-network product. The success of traditional insurers’ narrow network plans indicates that consumers are increasingly willing to trade off restrictions on provider choice for lower premiums, and the exchanges facilitate price competition by standardizing benefits across plans. Provider-owned insurers are poised to thrive in this environment, but only if they can translate theoretical efficiencies into real savings.

 Department of Health Policy and Management, Emory University (DHH), Atlanta, GA; USC Price School of Public Policy, University of Southern California (ET), Los Angeles, CA.

 This work was funded by Robert Wood Johnson Foundation grant 73800.

 The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article. 

 Concept and design (DHH, ET); acquisition of data (DHH, ET); drafting of the manuscript (DHH, ET); obtaining funding (DHH, ET). 

 David H. Howard, PhD, Department of Health Policy and Management, Emory University, 1518 Clifton Rd NE, Atlanta, GA 30030. E-mail: david.howard@emory.edu. 

1. Khanna G, Narula D, Rao D. The market evolution of provider-led health plans. McKinsey & Company website. http://healthcare.mckinsey.com/provider-led-health-plans-market-evolution. Published April 2016. Accessed June 15, 2017.

2. Federal Trade Commission and State of Illinois v. Advocate Health Care, Advocate Health and Hospitals Corporation, and Northshore University Healthcare System [redacted memorandum opinion and order]. Scribd website. https://www.scribd.com/document/342109735/Judge-Order#from_embed. Published March 16, 2017. Accessed June 15, 2017.

3. Newhouse, Joseph P., and the Insurance Experiment Group. 

Free for All? Lessons From the RAND Health Insurance Experiment. 

Cambridge, MA: Harvard University Press; 1993.

4. Frakt AB, Pizer SD, Feldman R. Plan-provider integration, premiums, and quality in the Medicare advantage market. 

. 2013;48(6, pt 1):1996-2013. doi: 10.1111/1475-6773.12076.

Articles

COMMENTARY

Provider-owned insurers have the potential to reduce costs, but face challenges building competitive provider networks.

Provider-Owned Insurers

 Statins are recommended to reduce serum cholesterol in patients at risk for atherosclerotic cardiovascular disease. Despite the prevalence of statin use, little is known about the indications for new prescriptions. We assessed the concordance of new statin prescriptions in the Veterans Health Administration (VHA) compared with the Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III, or ATP III) guidelines (the guidelines in force in 2012) and the American College of Cardiology (ACC)—American Heart Association (AHA) 2013 guidelines.

 We identified every patient who received a new prescription (no statin use in the prior year) in the VHA in 2012. Patients were excluded if they had incomplete data, triglycerides greater than 400 mg/dL, or fewer than 2 primary care visits to ensure adequate baseline data to calculate Framingham and ACC-AHA 2013 risk scores. 

We identified 250,243 new statin prescriptions in 2012 in the VHA, with 121,081 meeting inclusion criteria. Among new prescriptions, 68% were prescribed for primary prevention and 32% were prescribed for secondary prevention. Among patients receiving new statins for primary prevention, 48% did not have an indication supported by the ATP III guideline and 20% did not have an indication supported by the ACC/AHA guideline. Overall, approximately 19% of patients may have received a statin for an indication not supported by either guideline.

Veterans are commonly prescribed statins for indications not supported by professional society guidelines. The finding of common use of statins outside established guidelines represents an opportunity to improve the quality and value of the healthcare delivery.

New statin prescriptions at the Veterans Health Administration were reviewed using a cross-sectional study design. We found that statins are often prescribed for indications not supported by either the Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) or the American College of Cardiology—American Heart Association 2013 guidelines. 

Despite being commonly prescribed, little is known about indications for new statin prescriptions. 

Veterans are commonly prescribed statins for indications not supported by professional society guidelines. 

The identification of the use of statins outside established guidelines represents an opportunity to improve the quality and value of healthcare delivery.

Statins are recommended for the lowering of serum cholesterol levels in individuals at risk for, or who have, atherosclerotic cardiovascular disease (ASCVD). The Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III [ATP III]) has been issuing guidelines since 1988, and its last report was in 2004.

 The American College of Cardiology (ACC)—American Heart Association (AHA) updated the ATP guidelines in 2013.

Compared with ATP III, the ACC-AHA guidelines focused on evidence-based recommendations and expanded the number of people recommended to receive statin therapy. The greatest increase was among individuals without known ASCVD with lower levels of cardiovascular risk who would be taking statins for primary prevention.

 This expansion of the population recommended for statin prescriptions occurred at a time of controversy about the benefits of statins for primary prevention and during the focus on the elimination of unnecessary procedures, tests, and medications through the Choosing Wisely Campaign.

 The Veterans Health Administration (VHA), in particular, has also made a commitment to reduce misuse and overuse. 

Several studies have shown undertreatment of patients recommended to receive statins by both the ATP III and ACC/AHA guidelines.

 Prior examination of statin use in countries outside the United States suggests that overuse is also likely high.

 Much less is known about the indications for new prescriptions and the concordance of statin prescriptions to guidelines, with only 1 small single-center study from 2001 that found high rates of overuse.

To better understand the characteristics of patients receiving a new statin prescription, we set out to assess the concordance of new statin prescriptions in the VHA compared with the ATP III guidelines (the guidelines in force in 2012) and the ACC-AHA guidelines. 

We performed a national retrospective study of new statin prescriptions in the VHA in 2012 and examined concordance with the ATP III and ACC-AHA guidelines.

The VHA Pharmacy Benefits Management (PBM) database collects information on inpatient and outpatient medications dispensed at the VHA. The Corporate Data Warehouse (CDW) non-VHA medications database includes provider-entered information about medications provided to patients outside the VHA. Data on comorbidities, habits (eg, tobacco abuse), laboratory data, and vital statistics were obtained from the CDW, MedSAS, Decision Support System, and VHA Fee Basis files.

All patients 18 years or older who received any statin prescription at the VHA in 2012 were identified using the PBM database. Statins included lovastatin, simvastatin, fluvastatin, atorvastatin, pravastatin, rosuvastatin, and pitavastatin. At this stage, the identified cohort included both patients with new statin prescriptions and patients with longstanding statin prescriptions. Next, to identify new users, patients were excluded if they had been on any statin dispensed in the VHA in the prior 12 months. Patients were also excluded if they had ever, according to data from the CDW database, received a statin outside the VHA. Patients who were on any other cholesterol agent including gemfibrozil, fenofibrate, nicotinic acid, cholestyramine, colestipol, colesevelam, ezetimibe, neomycin, or probucol in the prior 12 months were also excluded to ensure an accurate assessment of each patient’s baseline lipid profile. 

Patients were further excluded if they had fewer than 2 primary care visits in the 12 months prior to the index prescription, because we wanted to focus on patients receiving primary care in the VHA and because these patients did not have enough baseline data for characterization. Patients were also excluded if they either did not have a complete lipid panel in the 6 months prior to entering the cohort or had triglycerides over 400 mg/dL, to ensure that patients did not have an alternative indication for a statin as a result of elevated triglycerides. 

Individuals were classified as receiving statins for secondary prevention if they had coronary artery disease (CAD), angina, stroke, transient ischemic attack (TIA), or peripheral artery disease (PAD). The 

International Classification of Disease, Ninth Revision, Clinical Modification

) and Current Procedural Terminology (CPT) were used to identify patients with CAD and PAD in the prior 10 years. A high-sensitivity stroke algorithm developed for VHA administrative data was used to identify patients with a history of stroke and TIA.

 The presence of diabetes and hypertension in the 2 years prior to receipt of statins was identified using a combination of receipt of medications and 

 The presence of other vascular risk factors, including carotid disease and abdominal aortic aneurysm, was identified using a combination of CPT codes and 

 codes, CPT codes, and medications used to construct these comorbidities are provided in the 

]. We used several sources of data to assess current smoking status. Patients were considered current tobacco smokers if they had evidence of current smoking from the electronic clinical reminders data (VHA health factors file), consults related to smoking services, or outpatient clinic visits related to smoking cessation programs in the year prior to receiving a statin.

We used age, sex, race, systolic blood pressure (SBP), presence of diabetes, use of blood pressure medication, total cholesterol, high-density lipoprotein cholesterol, and smoking status to calculate 10-year ASCVD risk at time of statin initiation using both the Framingham risk calculator (used in ATP III) and the ACC-AHA 2013 cardiovascular risk calculator.

 Patients were classified as black, white, or unknown race. Thirty percent of patients had an unknown race and were categorized as white for the purpose of ASCVD risk calculation. The ACC-AHA calculator was developed for patients aged 40 to 79 years. Among our cohort, 10% of patients were older than 79 years. We classified these patients as being aged 79 years for the purpose of calculating a 10-year risk score. If more than 1 SBP measurement was available, an average of the 2 most recent values was used. If more than 1 lipid panel had been performed, the most recent in terms of time to the statin prescription was used. Smoking status was determined as described above. 

Eligibility for statin therapy for primary prevention was assessed using the ATP III and ACC-AHA guidelines (

), using previously described hierarchical criteria.

 Patients were classified as receiving a statin for secondary prevention if they had a history of CAD, TIA, PAD, or stroke.

Data analysis was descriptive in nature with means and standard deviations reported (R version 3.1, R Development Core Team). Patients who received statins for primary and secondary prevention were described. The patients who received a statin for primary prevention were further classified as to whether they would have been eligible based on the ATP III guideline, ACC/AHA guideline, only the ATP III or ACC/AHA guideline but not both, or neither guideline. 

There were 2,125,908 patients at the VHA who received a statin in 2012, of whom 121,081 met the inclusion and exclusion criteria for analysis as a new statin prescription (

). Sixty-eight percent (82,600) of new statins were prescribed for primary prevention and 32% (38,481) for secondary prevention of ASCVD (

). Patients receiving statins for primary prevention were, on average, younger than patients receiving statins for known ASCVD, and they had a lower incidence of diabetes and hypertension. Among patients who received a new statin for primary prevention, 6% were younger than 40 years and 8% were older than 75 years.

Of the patients receiving new statins for primary prevention, 48% did not meet the ATP III guidelines and 20% did not fulfill the new ACC-AHA guidelines (

). Twenty-eight percent of patients met ACC-AHA guideline recommendations but not the ATP III guidelines, and only 0.3% of patients met the ATP III guidelines and not the ACC-AHA guidelines. About one-third (31%) of the patients recommended by only the ACC-AHA guidelines had low-density lipoprotein cholesterol (LDL-C) levels of 70 to 99 mg/dL. No patients older than 75 years were recommended by only the ATP III guidelines. Nineteen percent of new statin prescriptions were for patients who were not recommended by either guideline to receive statin therapy. About a fifth (19%) of these patients were younger than 40 years, and 19% had a baseline LDL-C level of less than 70 mg/dL. 

We found that, among veterans, the majority of new statins are prescribed for primary prevention. We found that a substantial proportion of new statin prescriptions would be considered outside the scope of current and former guideline recommendations. 

Approximately 20% of patients with an LDL-C less than 70 mg/dL or who were currently smoking were prescribed statins, despite not meeting either guideline recommendation. In these instances, it is possible that clinicians overestimated the risk of, for example, tobacco use alone, and prescribed a statin regardless of overall risk. Although guideline-discordant prescriptions appear relatively common in this sample, the challenges are that not all clinical scenarios can be included in the guidelines and that clinical judgment and clinician—patient shared decision making, including independent risk factor assessment, may explain a proportion of prescriptions outside guidelines. 

The rates of potential statin overuse appear high, but the possible harms to the patient from overuse of statins are difficult to assess and highly debated. Until recently, there was not a consensus definition of statin intolerance.

 The most commonly reported adverse effects (AEs) are muscle pains and weakness without elevations in creatinine kinase. The prevalence of these AEs in clinical practice is quite high (~20%), and these AEs are higher than those reported in randomized controlled trials.

 A high rate of statin discontinuation has been reported, including approximately 50% of patients who stop statins within the first year of prescription; intolerance has been suggested to be the primary source of discontinuation.

 Thus, the overall perception of negligible harms and the potential for a reduction in cardiovascular deaths may lead clinicians to overprescribe statins.

For some individuals currently taking statins without guideline-based indications, there may be no clear benefits, and there are potential unintended consequences of these prescriptions that should also be taken into account.

 For example, it has been documented that statin users experience more weight gain and become more sedentary than statin nonusers, possibly from the false sense of protection from taking a statin, leading individuals to ignore their diet and physical activity.

 In these patients, efforts that are focused on reduction of risks including exercise, weight loss, tobacco cessation, and management of hypertension, rather than initiation of a cholesterol medication, should be the primary focus.

Overprescribing statins certainly may also have financial implications for the health system, given the widespread use and the potentially long time horizon for younger patients prescribed statins for primary prevention. Currently, the “Choosing Wisely” campaign of the American Board of Internal Medicine Foundation, building on the work of the National Physicians Alliance, aims to list procedures, tests, and medications whose use should be questioned.

 With an estimated $750 billion annually spent on unnecessary and inefficient care, identification and reduction of potential overuse is imperative.

 However, de-implementation of statin use, even in populations not recommended by current guidelines, may be very challenging because of the perceived lack of harm and potential promise of long-term clinical benefit. Patients 40 years or younger may be committed to primary prevention for life. It is not at all clear whether such long-term use will be as effective or as harmless as is generally assumed. 

This study has several limitations. We initiated this study in 2013 using 2012 data and planned to use existing guidelines. The issuance of new guidelines by the AHA in 2013 complicated our study but also provided an opportunity to use both guidelines in assessing care. Second, our study is based on national VHA data and there may have been some under-recognition of risk factors, including family history used to calculate risk in the ATP III guidelines, and race used to determine risk in the ACC/AHA guidelines, because individuals of unknown race were assumed to be white. However, the VHA health record information is among the richest data sources available and we used previously developed algorithms, when available, to characterize patients.

 Third, we used VHA pharmacy data to ascertain new statin use. Some veterans may receive medications outside the VHA and may have been current users and not new users. In these patients, the cholesterol test may have been acquired while on a non-VHA statin, and therefore it would be incorrect to interpret the results as a baseline nonstatin value. We attempted to limit this possibility by requiring our cohort to have at least 2 VHA primary care visits prior to receipt of a statin prescription to largely confine our cohort to patients who receive primary care in the VHA. To assess how much non-VHA statin use affected our conclusions, we examined a random sample of 20 charts for patients whose receipt of statins was not concordant with either guideline. We did not identify any patients who received non-VA prescriptions in this group. To further evaluate non-VA statin use, we reviewed a random sample of 20 charts in patients with an LDL-C value of less than 70 mg/dL. We identified 4 patients (20%) in this select group who had evidence of non-VA statin use. These data suggest that the potential bias in the results related to missing data about non-VA medications is present but likely small in magnitude. Fourth, our sample is a predominantly male sample and may not be generalizable to statin use in the nonveteran US population. Finally, we have probably underestimated overuse of statins in the VHA because we did not examine statin use in populations that are unlikely to benefit, such as those who are receiving dialysis, have congestive heart failure, or have limited life expectancy. Overuse of statins among patients on established prescriptions—as opposed to new prescriptions&mdash;may be more common because there may be inertia related to stopping medications.

A major strength of the study is the population-based analysis of statin use in a national health system. By examining statin use across the entire VHA system, we gain an understanding of national prescribing patterns much more so than would be possible in smaller studies.

 By understanding national prescribing patterns, we can potentially identify high-value areas of quality improvement. Based on our study, increasing awareness among physicians about the lack of efficacy of statin use in young patients without ASCVD, or in patients with an LDL-C value of less than 70 mg/dL, might reduce inappropriate prescriptions and is a potential target to decrease overuse. 

Low-value statin use is common in the VHA. The finding of common use of statins outside established guidelines represents an opportunity to improve the quality and value of the healthcare delivery.

 Department of Medicine, Division of Cardiovascular Medicine, University of Michigan (TC), Ann Arbor, MI; San Francisco VA Medical Center (TC, MV, JB, EE, SK), San Francisco, CA; Department of Medicine, Division of Cardiology (RR) and Division of General Internal Medicine (SK), and Department of Epidemiology and Biostatistics (JB), University of California, San Francisco, San Francisco, CA; VHA HSR&D Center for Health Information and Communication, Richard L. Roudebush VA Medical Center (DMB), Indianapolis, IN; Department of Internal Medicine, Indiana University School of Medicine (DMB), Indianapolis, IN.

CLINICAL

New statin prescriptions at the Veterans Health Administration were reviewed using a cross-sectional study design. Statins were frequently prescribed outside of guideline recommendations.

Guideline Concordance of New Statin Prescriptions: Who Got a Statin?

To examine temporal trends and racial/ethnic differences in hip fracture incidence and mortality outcome in older men.

We ascertained men 50 years or older with a hip fracture during 2000 to 2010 in a diverse northern California healthcare population. Age, comorbidity index, hip fracture incidence, and all-cause mortality up to 12 months post fracture were examined and compared by race/ethnicity.

 A total of 6247 men (aged 79.3 ± 9.8 years) experienced a hip fracture during 2000 to 2010: 81.4% were white, 7.5% Hispanic, 3.8% black, and 3.9% Asian. The age-adjusted annual incidence of hip fracture averaged 127 per 100,000, ranging from 116 to 139 per 100,000 during this period. In 2010, the age-adjusted incidence of hip fracture was highest among white men (137), followed by Hispanic (98) and black (80), and was lowest among Asian men (45 per 100,000). Mortality following hip fracture was 11.1%, 19.8%, 25.4%, and 32.9%, within 1, 3, 6, and 12 months, respectively, and increased with age. One-year mortality was similar for whites (33.7%), blacks (32.4%), and Hispanics (31.1%), but lower for Asians (23.1%; 

<.05). Adjusting for age, comorbidity index, and calendar year, Asians remained at lower mortality risk compared with whites (adjusted odds ratio, 0.62; 95% confidence interval, 0.45-0.86). 

Although hip fracture rates were largely stable among older men, contemporary rates of hip fracture were highest for white and lowest for Asian men. One-year mortality was similar for white, black, and Hispanic men, but significantly lower for Asians. Future studies should investigate factors underlying observed ethnic differences in fracture outcome among US men.

Within an integrated healthcare delivery system, overall hip fracture rates were stable in older men, but important racial and ethnic differences in the incidence of hip fracture and subsequent 1-year mortality following hip fracture were observed. 

The incidence of hip fracture was highest among white men, lower among Hispanic and black men, and lowest among Asian men. 

Compared with white men and adjusting for age, comorbidity burden, and fracture year, Asian men experienced lower 1-year mortality risk following hip fracture, whereas differences were not significant for Hispanic and black men.

The economic burden of hip fractures in the United States is projected to exceed $17 billion by 2025.

 Although declines in hip fracture incidence are evident since the mid-1990s,

 the growth and demographic shifts of the aging US population are expected to contribute to an increasing number of hip fractures among older individuals and those of nonwhite race/ethnicity.

 Hip fracture rates are substantially higher in women compared with men,

 affecting whites more than blacks, Hispanics, and Asians.

 However, men experience greater post-fracture morbidity and mortality,

 even after adjustment for differences in age and prefracture comorbidity.

 Men also experience a higher incidence of postoperative complications, such as delirium, congestive heart failure, renal failure, decubitus ulcers, and infections, and are less likely to return to their functional baseline compared with women.

In 2014, the International Osteoporosis Foundation issued a report on osteoporosis in men, drawing attention to the increasing burden of osteoporosis and fragility fractures as the population of aging men increases, especially in Asia and Latin America.

 This report cited data from the US National Hospital Discharge Survey, estimating a 51.8% increase in hip fractures among men from 2010 to 2030, compared with an expected 3.5% decline in women.

 Few studies have examined contemporary trends within diverse integrated healthcare delivery settings. Racial and ethnic population trends in men are also limited, but hip fracture rates appear to be declining among white men, although not necessarily among men of nonwhite race/ethnicity.

Large observational cohorts have established important racial/ethnic differences in mortality outcome following hip fracture, with the majority of data comprising outcomes reported among women.

 Historically, black women experience higher post-fracture mortality compared with white women, despite lower hip fracture rates.

 Within our healthcare system, we found similar mortality risk following hip fracture for black and white women, but lower mortality for Hispanic and Asian women.

 Less is known regarding racial/ethnic disparities in postfracture mortality among men, particularly Asians. This study examines contemporary rates of hip fracture and postfracture mortality in a diverse population of older men, using data from a large integrated healthcare delivery system.

The source population included health plan members of Kaiser Permanente Northern California (KPNC), an integrated healthcare system serving over 3.2 million members, with centralized electronic databases of all outpatient, hospitalization, and administrative records.  For these analyses, men 50 years or older with a principal hospital discharge diagnosis of proximal femur fracture (

International Classification of Diseases, 9th edition

] codes 820.0x, 820.2x, 820.8) during January 1, 2000, to December 31, 2010, were identified, examining the first qualifying fracture per calendar year. Open fractures (

: 820.1x, 820.3x, 820.9) and those with major trauma (secondary 

: E800-E848) were excluded. This study was approved by the Kaiser Foundation Research Institute Institutional Review Board. 

Demographic data were obtained from administrative databases, including self-reported race/ethnicity categorized as non-Hispanic white, non-Hispanic black, Hispanic, Asian, and other (including multiple race) or unknown race/ethnicity. The comorbidity index was calculated using Deyo and colleagues’ modification of the Charlson Comorbidity Index,

 based on diagnosis and procedure codes obtained from hospitalization, emergency, and ambulatory visits during the year prior to hip fracture. All-cause mortality was determined up to 12 months following hip fracture, using health plan and administrative databases (updated with information from state death certificates and Social Security Administration files), as previously described.

 Less than 1% of the study cohort had membership cessation within 1 year post fracture, excluding ascertained death outcomes; these individuals were presumed alive at 1 year.

Comparisons between subgroups were examined by Student’s 

 test for categorical variables. Hip fracture incidence was examined using age-eligible denominators for each calendar year, with rates adjusted for age using the 2010 US Census standard population data.

 Trends in age-adjusted fracture incidence were examined using generalized estimating equations, as previously described,

 with racial and ethnic differences examined for the final year of observation (2010). Multivariable logistic regression was conducted to examine the independent association of race/ethnicity and 1-year postfracture mortality, adjusting for age and comorbidity index. All analyses were conducted using SAS version 9.3 (SAS Institute; Cary, North Carolina). 

 <.05 was chosen as the criterion for statistical significance.

From 2000 to 2010, 6247 men 50 years or older experienced a qualifying hip fracture, with a mean age of 79.3 ± 9.8 years at the time of hip fracture; three-fourths (74.5%) were 75 years or older. The demographic distribution included 81.4% white, 7.5% Hispanic, 3.9% Asian, 3.8% black, and 3.6% other/unknown race/ethnicity, similar to KPNC women experiencing hip fracture.

 We found that age at fracture differed slightly by race/ethnicity, with black (mean age of 77.8 ± 9.8 years) and Asian (77.5 ± 10.6 years) men experiencing hip fracture at slightly younger ages compared with white men (79.6 ± 9.8 years; 

<.05), but no difference between Hispanic (78.7 ± 9.6 years) and white men.  

The average annual age-adjusted incidence of hip fracture during 2000 to 2010 was 127 per 100,000 men, ranging from 116 to 139 per 100,000, although no overall significant trend was observed (

). Racial/ethnic differences in contemporary hip fracture rates were evident.  In 2010, the age-adjusted incidence of hip fracture was highest for white men (137 per 100,000), similar among Hispanic (98 per 100,000) and black (80 per 100,000) men, and lowest in Asian (45 per 100,000) men (

<.01 for all race groups vs white and for Asian vs Hispanic).

Of the 6247 men who experienced a hip fracture during 2000 to 2010, nearly one-third died during the ensuing year. All-cause mortality rates were 11.1%, 19.8%, 25.4%, and 32.9% at 1, 3, 6, and 12 months following hip fracture, respectively. These rates are 1.5-fold higher than mortality rates reported in KPNC women post hip fracture (17.0% at 6 months, 22.8% at 12 months).

 As expected, mortality following hip fracture increased substantially with age (

 [a]), with lower 1-year mortality in younger, compared with older, men post hip fracture across all age group comparisons (

We further examined postfracture mortality rates by race/ethnicity (Figure 2 [b]).  Racial and ethnic differences at 1 year (

= .02) were notable for lower 1-year mortality among Asian men (23.1%; 

<.05) compared with white (33.7%), black (32.4%), and Hispanic (31.1%) men.  However, Asian men were also slightly younger at the time of hip fracture. Adjusting for age, Asian men remained at significantly lower mortality risk compared with white men (adjusted odds ratio [aOR], 0.65; 95% confidence interval [CI], 0.47-0.88), whereas no differences were seen for Hispanic (aOR, 0.93; 95% CI, 0.76-1.15) and black (aOR, 1.05; 95% CI, 0.79-1.15) men compared with white men at 1 year. These results were largely unchanged after additionally adjusting for baseline comorbidity index and calendar year of fracture (aOR, 0.62; 95% CI, 0.45-0.86 for Asian; aOR, 0.88; 95% CI, 0.71-1.09 for Hispanic; aOR, 0.96; 95% CI, 0.71-1.29 for black compared with white men).

Hip fracture remains a major public health issue, contributing substantially to healthcare cost

 and morbidity and mortality of older individuals. Screening guidelines for osteoporosis, a major risk factor for hip fracture, have been well established for older women,

 and guidelines are now in place for older men.

 However, according to the US Preventive Services Task Force, there is insufficient evidence to evaluate the risks and benefits of osteoporosis screening in men,

 and men are much less likely than women to receive treatment for osteoporosis.

 Even after hip fracture, only a fraction of men receive osteoporosis treatment compared with women.

 Within our health plan, the annual incidence of hip fracture among older men averaged 127 per 100,000, with an incidence of 116 per 100,000 during the final year of observation (2010). These data provide a benchmark for future assessment of fracture trends following implementation of osteoporosis outreach programs for men, similar to earlier outreach programs for women contributing to reductions in hip fracture and subsequent mortality risk.

 Kaiser Permanente Southern California reported temporal reductions in hip fracture rates for both men and women,

 attributed to regionwide efforts and establishment of a multidisciplinary osteoporosis management program to aggressively screen and treat patients for osteoporosis.

Given the growing US population, Hispanics and Asians are projected to have the greatest rate of increase in hip fracture costs by 2025,

 underscoring the need to better understand potential disparities in fracture incidence and mortality. In our study, hip fracture rates were highest in white men and lower for other races and ethnicities—similar to findings in women

 Racial and ethnic differences in bone mineral density (BMD) are known, with blacks having higher BMD compared with whites,

 accounting in part for lower fracture risk.

 Smaller body size contributes to lower BMD in Asian men,

 although BMD is not the only measure of femur strength given that Asian men also have a lower incidence of fracture compared with white counterparts.

  Differences in femur size, hip geometry (shorter hip axis length in Asians), and/or hip strength indices may also explain ethnic variation in fracture risk,

 as well as cultural influences on activity level and fall risk. Within our health plan, 10% of men 65 years or older are Asian,

 whereas only 3.9% of men with hip fracture are Asian. A two-thirds lower age-adjusted incidence of hip fracture was observed in Asian compared with white men.  

Although men contribute a much smaller subset of the hip fracture population than women, they suffer greater morbidity and mortality compared with women. In our study, 1-year mortality following hip fracture was 32.9% in men—substantially higher than women (22.8%) within the same healthcare setting.

 National and regional data also demonstrate higher mortality rates up to 1 year post hip fracture in men compared with women.

 Because men may be less prone to falls, those who do experience hip fracture may have increased frailty, greater comorbidity, higher susceptibility to medical complications (eg, pneumonia), and potentially lower rates of subsequent bisphosphonate therapy compared with women.

 Whether the observed gender differences in mortality risk following hip fracture relate to pre- or postfracture health, including a potentially greater contribution of infection-related complications in men,

Our study contributes to the growing recognition of racial/ethnic differences in postfracture mortality. Similar to women, 1-year mortality rates following hip fracture were lower in Asian men, but mortality rates were comparable for white, black, and Hispanic men. A recent study examining nonfederal hospital admissions for hip fracture—related procedures in California reported more than one-third lower postfracture mortality risk for Asian men and somewhat lower mortality risk among black and Hispanic compared with white men.

 Whether hospital readmission or other postfracture complications,

 as well as ethnic differences in family structure, social support, and use of rehabilitation services,

 contribute to these findings is unclear. Contemporary data pertaining to US Asian men also remain limited. Collectively, these findings emphasize the need to further investigate factors underlying the observed ethnic differences in hip fracture outcomes among older men.

Our study has several limitations. First, we did not examine specific preexisting comorbidities, functional status, osteoporosis risk factors, and treatment, which may differ by race and play an important role in patient outcomes. Second, information on mobility and/or functional independence, discharge to inpatient rehabilitation, body mass index, and nutritional status were not systematically available. Lastly, we were unable to account for population trends in mortality to determine whether the observed racial differences reflect mortality patterns within our health plan population, irrespective of hip fracture. 

The strengths of our study include access to an extremely large and diverse population, with comprehensive data on mortality outcomes and hospitalized events. Asians represent one of the fastest growing ethnic subgroups within the United States, among whom a better understanding of hip fracture epidemiology and outcome has become increasingly important. These data are among the first to examine contemporary differences in both hip fracture rates and mortality outcome in men of Asian ethnicity and white race receiving care within the same healthcare delivery system.

We noted important racial/ethnic differences in hip fracture incidence and 1-year mortality outcome following the fracture. Compared with men of white race, Asian men had two-thirds lower hip fracture incidence and one-third lower mortality risk at 1 year following hip fracture. As the aging population becomes increasingly diverse, a greater understanding of the cultural, social, and health-related factors affecting fracture outcomes in healthcare settings will optimize the targeting of multidisciplinary efforts to reduce morbidity and mortality following hip fracture in men.

 Department of Medicine, Kaiser Permanente Oakland Medical Center (LHL, JCL), Oakland, CA; Division of Research, Kaiser Permanente Northern California (MC, JRG, JCL), Oakland, CA.

 Ms Chandra has previously received research funding from Amgen. Dr Lo has previously received research funding from Amgen and Sanofi. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article. 

 Concept and design (LHL, MC, JCL); acquisition of data (MC, JCL); analysis and interpretation of data (LHL, MC, JRG, JCL); drafting of the manuscript (LHL, JCL); critical revision of the manuscript for important intellectual content (LHL, MC, JRG, JCL); statistical analysis (MC); provision of patients or study materials (JCL); administrative, technical, or logistic support (LHL, JRG, JCL); and supervision (JCL). 

 Joan C. Lo, MD, Kaiser Permanente Northern California, 2000 Broadway, Oakland, CA 94612. E-mail: Joan.C.Lo@kp.org. 

1. Burge R, Dawson-Hughes B, Solomon DH, Wong JB, King A, Tosteson A. Incidence and economic burden of osteoporosis-related fractures in the United States, 2005-2025. 

2007;22(3):465-475. doi: 10.1359/jbmr.061113.

2. Brauer CA, Coca-Perraillon M, Cutler DM, Rosen AB. Incidence and mortality of hip fractures in the United States. 

2009;302(14):1573-1579. doi: 10.1001/jama.2009.1462.

3. Orces CH. In-hospital hip fracture mortality trends in older adults: the National Hospital Discharge Survey, 1988-2007. 

2013;61(12):2248-2249. doi: 10.1111/jgs.12567.

4. Stevens JA, Rudd RA. The impact of decreasing U.S. hip fracture rates on future hip fracture estimates. 

2013;24(10):2725-2728. doi: 10.1007/s00198-013-2375-9.

5. Zingmond DS, Melton LJ 3rd, Silverman SL. Increasing hip fracture incidence in California Hispanics, 1983 to 2000. 

2004;15(8):603-610. doi: 10.1007/s00198-004-1592-7.

6. Jacobsen SJ, Goldberg J, Miles TP, Brody JA, Stiers W, Rimm AA. Race and sex differences in mortality following fracture of the hip. 

7. Lo JC, Zheng P, Grimsrud CD, et al. Racial/ethnic differences in hip and diaphyseal femur fractures. 

2014;25(9):2313-2318. doi: 10.1007/s00198-014-2750-1.

8. Pothiwala P, Evans EM, Chapman-Novakofski KM. Ethnic variation in risk for osteoporosis among women: a review of biological and behavioral factors. 

2006;15(6):709-719. doi: 10.1089/jwh.2006.15.709.

9. Endo Y, Aharonoff GB, Zuckerman JD, Egol KA, Koval KJ. Gender differences in patients with hip fracture: a greater risk of morbidity and mortality in men. 

10. Penrod JD, Litke A, Hawkes WG, et al. The association of race, gender, and comorbidity with mortality and function after hip fracture. 

11. Kannegaard PN, van der Mark S, Eiken P, Abrahamsen B. Excess mortality in men compared with women following a hip fracture. National analysis of comedications, comorbidity and survival. 

2010;39(2):203-209. doi: 10.1093/ageing/afp221.

12. Haentjens P, Magaziner J, Colón-Emeric CS, et al. Meta-analysis: excess mortality after hip fracture among older women and men. 

2010;152(6):380-390. doi: 10.7326/0003-4819-152-6-201003160-00008.

13. Forsén L, Sogaard AJ, Meyer HE, Edna T, Kopjar B. Survival after hip fracture: short- and long-term excess mortality according to age and gender. 

14. Sterling RS. Gender and race/ethnicity differences in hip fracture incidence, morbidity, mortality, and function. 

2011;469(7):1913-1918. doi: 10.1007/s11999-010-1736-3.

15. Wehren LE, Hawkes WG, Orwig DL, Hebel JR, Zimmerman SI, Magaziner J. Gender differences in mortality after hip fracture: the role of infection. 

2003;18(12):2231-2237. doi: 10.1359/jbmr.2003.18.12.2231.

TRENDS FROM THE FIELD

Racial and ethnic differences in hip fracture incidence and mortality outcome were observed within a diverse population of older men, with lower rates of both among Asians.

September 2017

Provider-Owned Insurers

Guideline Concordance of New Statin Prescriptions: Who Got a Statin?

Racial and Ethnic Differences in Hip Fracture Outcomes in Men

In-Gap Discounts in Medicare Part D and Specialty Drug Use

The Effect of Narrow Network Plans on Out-of-Pocket Cost

Integrating Behavioral Health Under an ACO Global Budget: Barriers and Progress in Oregon

Evaluation of a Packaging Approach to Improve Cholesterol Medication Adherence

Treatment Barriers Among Younger and Older Socioeconomically Disadvantaged Smokers

Against the Current: Back-Transfer as a Mechanism for Rural Regionalization

Association Between FDA Black Box Warnings and Medicare Formulary Coverage Changes