 To test the validity of using administrative data to identify patients with osteoporosis or low bone mineral density (BMD) and high risk for osteoporotic fractures.

 We conducted a retrospective cohort study.

We analyzed data from a managed care plan in Massachusetts. We developed 6 case-identification algorithms based on number of osteoporosis (OP) diagnoses, clinical setting of the OP diagnosis, timing of the OP diagnosis relative to BMD test, and clinical fracture risk factors adapted from the World Health Organization Fracture Risk Assessment Tool. We validated the algorithms against BMD results and calculated sensitivity, specificity, and positive predictive value (PPV) against 2 diagnostic criteria (T-score 

 When compared against the first criterion (T-score

—2.5), the sensitivity of algorithm (35% to 80%), specificity (65% to 93%), PPV (44% to 63%), and adding fracture risk factors did not improve case identification. When compared against the expanded criterion (T-score

—2.0), we found the sensitivity of the algorithms ranged from 23% to 63%, specificity from 72% to 95%, and PPV from 67% to 83%. Including fracture risk in the expanded OP criterion improved case identification, and the algorithms achieved the highest PPV: 70% to 85%.

Identifying patients with OP or low BMD and high risk for osteoporotic fractures ispossible in administrative data if using information about both OP diagnoses and fracture risk profile.

		Our algorithm uniquely identifies patients with osteoporosis (OP) and low bone mineral density (BMD) and does not require prior knowledge of OP prescriptions. Therefore it is useful in identifying untreated patients who could benefit from OP treatment.

		 Incorporating the Fracture Risk Assessment Tool clinical risk factors in case-identification algorithms may be a useful strategy to identify patients with OP or low BMD.

		 Further studies are warranted to understand the under-documentation of OP.

Case identification of patients at risk for an osteoporotic fracture remains a challenging problem, especially when using administrative data. Approximately 50% of osteoporosis (OP) is undiagnosed or undertreated, and testing for low bone mineral density (BMD) is often not conducted even after a fracture.

 For instance, over 20% of adults 65 years or older never received a BMD test or diagnosis of OP after experiencing an osteoporotic fracture, as captured in 7 years of follow-up Medicare claims data.

 A diagnosis of OP may also be underrepresented in administrative data since secondary conditions are less likely to be treated and recorded when patients have multiple chronic medical diseases.

 Furthermore, osteoporotic fracture risk encompasses more clinical factors than just the diagnosis of OP and many of these factors are not typically captured in administrative data. Weight, likelihood of falling, low BMD, family history of fracture, smoking, and alcohol use are significant predictors of osteoporotic fracture yet are unmeasured confounders in claims-based studies. As a result, characterizing patients in administrative data by their OP diagnosis and clinical risk profile for osteoporotic fracture is susceptible to misclassification error.

Despite these limitations, administrative data remain a convenient and important source for studying patient populations who might be suitable candidates for receiving OP prescriptions to prevent osteoporotic fracture. The challenge is to create a valid approach for correctly identifying the at-risk patient population, given the data limitations described above. Only 3 studies have addressed this issue in administrative data and each has limitations. One study found that a combination of diagnostic information available in administrative data and fracture claims may be used to identify individuals with OP, although that strategy has not been validated against BMD criteria.

 The other 2 studies found that prescriptions for OP drugs are a reliable marker of OP

but only after using additional BMD data to identify patients with osteopenia and exclude them from the study sample.

 Both studies could not be used for identifying untreated patients since they used receipt of OP prescriptions in their case-identification algorithms.

The goal of the present study is to create and validate case-identification algorithms suitable for finding patients with OP or low BMD who may require OP prescriptions. The algorithm must use only data elements commonly found in administrative data, yet not incorporate the receipt of OP prescriptions. Furthermore, the algorithm must incorporate clinical risk factors for OP fracture as defined by the World Health Organization (WHO) Fracture Risk Assessment (FRAX) Tool clinical risk factors (CRFs).

We used claims data for approximately 100,000 members of a managed care plan who received their care from a multispecialty group practice in central Massachusetts (between January 1, 2008, and December 31, 2009). The database captures information on demographics, enrollment, outpatient encounters, hospitalizations, diagnostic and therapeutic procedures, and dispensings of prescription medications. The claims data were linked to a clinical data set that contained results of BMD testing.

We first selected women who received a BMD test or had a diagnosis of OP (n = 5340). We included women 50 years or older who had continuous enrollment in the managed care plan for 2 years and who had a BMD test in 2009. Since the BMD test is not usually conducted every year, a patient with a BMD test in 2009 is not likely to have another BMD in 2008. We excluded women who had less than 1 year of enrollment before the BMD test.

The BMD was examined using the dual energy x-ray absorptiometry (DXA) scan (Hologic, Waltham, Massachusetts) performed in the group practice’s facility. The standard to make the diagnosis was based on the T-score at the lumbar spine, femoral neck, or total hip. We first adopted the WHO criterion, T-score 

—2.5, as the gold standard to define OP.

 However, in actual clinical practice, treatment decisions are based not only on T-score but also risk factors for osteoporotic fracture.

For instance, a physician may prescribe OP treatment for a woman with a T-score of —2.0 if she presents with 1 or more osteoporotic fracture risk factors. Therefore, we also applied an expanded diagnostic criterion, T-score 

We identified clinical risk factors for osteoporotic fracture based on the WHO FRAX tool.

 The FRAX tool, introduced in 2008, calculates the 10-year probability of a major osteoporotic fracture and of a hip fracture using easily obtained CRFs and bone density information when available. 

 The CRFs in the FRAX tool include weight, height, history of fracture, parental history of hip fracture, current tobacco smoking, ever long-term use of an oral glucocorticoid, rheumatoid arthritis (RA), other cause of secondary OP, and daily alcohol consumption of 3 or more units. Most administrative data sets do not include height, weight, or family history. However, we were able to identify 5 CRFs or proxies of CRFs including history of an osteoporotic fracture, use of an oral glucocorticoid medication, and diagnosis of RA. We used diagnosis of chronic obstructive pulmonary disease (COPD) as a proxy for smoking, and diagnosis of alcoholism as a proxy for current daily consumption of 3 or more units of alcohol.

Only risk factors identified during the period before the BMD test (average 1.5 years) were included. History of an osteoporotic fracture was defined as having a fracture of the wrist, hip, or proximal humerus and was identified using 

International Classification of Diseases, Ninth Revision, Clinical Modification

) and CPT codes, as developed by previous research.

 Glucocorticoid use was defined as receiving oral prednisone, methylprednisolone, hydrocortisone, cortisone, prednisolone, triamcinolone, dexamethasone, or budesonide for at least 90 days. We included oral glucocorticoids identified using the national drug code. A patient was considered as having RA if she had a diagnostic code of 714.xx. Smoking status was not readily identified from administrative data. Therefore, we adopted a previously used method that used COPD diagnosis as a proxy for smoking

 and COPD was identified using diagnostic codes.

 for alcohol dependence syndrome and 305.0x for nondependent alcohol abuse.

Algorithms for OP Classification and Possible Error Correction

We developed 6 case-identification algorithms (

) that included number of OP diagnoses, setting of OP diagnosis (inpatient or outpatient), and timing of OP diagnosis relative to BMD test. The first 3 algorithms were based on diagnostic codes of OP from outpatient encounters and hospital claims: (1) at least 1

 code for OP in outpatient encounters or hospital claims, (2) at least 1 code for OP in hospital claims or at least 2 in outpatient encounter claims, and (3) at least 2 codes for OP regardless of setting (Algorithms 1 to 3, Table 1).

We then repeated the 3 algorithms above for the sample after possible error correction (Algorithms 4-6, Table 1). A patient may be tentatively diagnosed as having OP as a justification for a BMD test, but the BMD test result may actually rule out this diagnosis. Therefore, we corrected for this possible error by removing the OP diagnosis if (1) all of the

codes for OP appeared before the BMD test, and (2) the patient had at least 3 months’ enrollment after the BMD test.

We described the distribution of patient characteristics stratified by BMD test results (T-score >—2.0, –2.5 < T-score 

—2.5). We compared means using analysis of variance (ANOVA) and proportions using 

 test or Fisher’s exact test. We validated the performance of our OP classification algorithms against the WHO standard, T-score —2.5 or lower, and then an expanded diagnostic criterion, T-score –2.0 or lower. We calculated sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the receiver operating characteristic curve (AUC) for each algorithm. Finally, we applied the same algorithms to the high-risk population, patients with at least 1 of the 5 FRAX risk factors that we were able to identify. Analyses were performed with SAS 9.2 (SAS Institute, Cary, North Carolina).

This study was approved by the Saint Vincent Hospital/ Fallon Clinic/Fallon Community Health Plan Institutional Review Board, Worcester, MA.

There were 5340 patients who had at least 1

 diagnosis of OP or BMD test results between January 1, 2008, and December 31, 2009. We excluded 733 patients who were not continuously enrolled. We further excluded 1987 patients who did not receive a DXA test. The study sample included 2520 patients (

describes characteristics for the study population (n = 2520) stratified by BMD test results. We found that 25.7% (n = 647) had a Tscore

—2.5 at 1 or more sites measured (lumbar spine, femoral neck, or total hip) and 25.0% (n = 631) had at least 1 of the FRAX risk factors for osteoporotic fracture as measured in our study. Compared with women with T-score greater than –2.5, those with T-score less than or equal to –2.5 were older (T-score 

—2.5 vs T-score between –2.5 and –2.0; vs T-score >–2.0; age 74.3 vs 71.2 vs 67.6 years; 

 <.001). Women with T-scores less than or equal to —2.5 also had more fracture risk factors identified in administrative data. For instance, about 29.5% of women with a T-score less than or equal to –2.5 had more than 1 risk factor compared with 25.5% and 22.4% for women with a T-score from –2.5 to –2.0 and women with a T-score greater than –2.0, respectively.

 shows the sensitivity, specificity, PPV, NPV, and AUC of each algorithm when the diagnostic criterion was a T-score less than or equal to —2.5 (scenarios 1 and 2). We found in scenario 1 (the total study population, n = 2520) that sensitivity varied from 34.9% (algorithm 6) to 80.4% (algorithm 1); specificity from 65.3% (algorithm 1) to 92.8% (algorithms 5 and 6); PPV from 43.1% (algorithm 4) to 63% (algorithm 5); and AUC from about 0.636 (algorithm 4) to 0.737 (algorithm 2). After restricting to women with at least 1 FRAX risk factor (scenario 2, n = 631), sensitivity, specificity, PPV, NPV, and AUC changed only slightly. Sensitivity ranged from 38.7% to 83.2%; specificity from 59.1% to 90.5%; PPV from 46.9% to 64.1%; and AUC from 0.646 to 0.718.

 shows the sensitivity, specificity, PPV, and NPV when the diagnostic criterion was a T-score less than or equal to —2.0 (scenarios 3 and 4). Compared with the statistics in scenario 1, sensitivity (ranging from 22.8% to 63.1%) decreased; specificity changed slightly (ranging from 71.5% to 95%); and PPV improved substantially, varying from 66.5% to 83.1% in scenario 3 (the total study population, n = 2520). After restricting to the population with at least 1 of the 5 FRAX risk factors we identified (scenario 4), we found that sensitivity was still lower than that in scenario 1 and specificity changed slightly. However, the PPV was the highest among all 4 scenarios, ranging from 70.6% to 85.0%.

To our knowledge, our study is the first to test the validity of identifying patients who have OP or who have low BMD (osteopenia) and additional risk factors for osteoporotic fractures, using algorithms applied to administrative data in a US population. We used the results of BMD tests as the diagnostic standard and assessed the performance of our algorithms using a strict OP criterion (T-score 

—2.5) and an expanded criterion (T-score 

—2.0) in a population with at least 1 of the 5 FRAX risk factors we were able to identify in administrative data. Our work is also unique in that we did not require information about use of OP medications, and thus our methods are suitable for identifying the at-risk patient population before OP intervention. However, it is possible prescription claims are the only source to identify OP for some patients. In our study, we found that 489 patients had a bisphosphonate prescription (alendronate or risedronate) before the BMD test. We conducted a sensitivity analysis incorporating 2 commonly used bisphosphonates, alendronate and risedronate, into our algorithms and found the results of case-identification remained similar (data not shown).

Our study is also unique in that we incorporated 5 CRFs or proxies of CRFs from the WHO FRAX tool (RA, previous fracture, alcohol use, smoking, and history of treatment with oral glucocorticoids) into the risk assessment. This expansion incorporates current clinical recommendations that treatment to prevent osteoporotic fractures may also be appropriate for patients with a T-score better than —2.5 if other risk factors are present.

 In fact, we found that FRAX information did improve predictive power when the diagnostic criterion was a T-score less than or equal to —2.0, but not for the scenarios where the diagnostic criterion was a T-score less than or equal to –2.5. This suggests that incorporating FRAX risk factors may improve more of the performance of case-identification algorithms for osteopenia rather than OP using administrative data. Furthermore, other predictors, such as age, may improve the performance of the study algorithms because older age is= associated with OP. However, when we stratified the sample by age (

65 or <65 years of age), we found the algorithms performed similarly between the stratified sample and unstratified sample. Therefore, we report the results without stratification. A possible reason for this observation is that we have an older population with median age of 70 years and 25th percentile of 63 years.

Articles

WEB EXCLUSIVE

Identifying patients with osteoporosis or low bone mineral density and high risk for osteoporotic fractures is possible in administrative data using osteoporosis diagnoses and fracture risk profile.

Identifying Patients With Osteoporosis or at Risk for Osteoporotic Fractures

To assess the feasibility of engaging adults with diabetes in self management behaviors between clinic visits by using cell phone text messaging to provide blood sugar measurement prompts and appointment reminders.

 Quasi-experimental pilot among adult diabetic patients with cell phones who receive regular care at a federally qualified community health center in Denver, Colorado, which serves a population that is predominantly either uninsured (41%) or on Medicaid or Medicare (56%).

Patients (N = 47) received text message prompts over a 3-month period. Blood sugar readings were requested 3 times per week (Monday, Wednesday, and Friday). Reminders were sent 7, 3, and 1 day(s) before each scheduled appointment. Acknowledgments were returned for all patient-sent messages. Focus groups were conducted in English and Spanish with selected patients (n = 8).

 Patients of all ages were active participants. Correctly formatted responses were received for 67.3% of 1585 prompts. More than three-fourths (79%) of the cohort responded to more than 50% of their prompts. The appointment analysis was underpowered to detect significant changes in attendance. Participants reported increased social support, feelings that the program “made them accountable,” and increased awareness of health information. Two-thirds (66%) of patients provided glucose readings when prompted during the study, compared with 12% at 2 preceding clinic visits.

 For certain patients, cell phone—based text messaging may enhance chronic disease management support and patient-provider communications beyond the clinic setting.

Healthcare systems can better support chronic disease management by offering patients customized communication that identifies and supports patients outside the clinic visit setting.

Health information technology (HIT) can be used to provide needed support to patients for chronic disease management.

In this study, blood sugar measurement prompts and appointment reminders were provided to adult patients with diabetes through cell phone text messaging to promote engagement in self-care behaviors.

Low-cost, mobile HIT such as that used in this study may feasibly enhance regular patient-provider communications and engagement with the healthcare system for some patients.

The impact of chronic disease is substantial. An estimated 60% of deaths worldwide are thought to be caused by chronic disease,

 with 4 chronic diseases—cardiovascular disease, cancer, chronic respiratory disease, and diabetes&mdash;responsible for 29 million deaths in 2002 alone.

 The situation is no better in the United States, where 70% of deaths annually are attributable to chronic disease

 and almost 50% of the population has at least 1 chronic illness.

Patients need to manage chronic diseases on a daily basis, yet US healthcare systems are not well structured to support self management across large patient populations.

 Both patients and providers express frustration with the standard approach to managing chronic disease through the 20-minute provider-driven clinic visit.

 In addition, traditionally vulnerable and medically underserved groups, which are more likely to have poorly controlled chronic illness,

 are also more likely to experience barriers in accessing healthcare, which means that chronic disease management strategies focused on only during the clinic visit present further challenges for these patients.

Health information technology (HIT) has been shown to be useful in helping patients to manage chronic conditions outside the clinic setting.

 The combination of Web-based health informatics with case management strategies has been reported to improve blood pressure control among patients with hypertension

 and glycemic control among patients with diabetes.

 Cell phone—based text messaging has also been associated with improved glycemic control when used to assist with diabetes case management.

 The latter is of particular interest given that high rates of cell phone access are reported among groups in the United States that exhibit low rates of computer and Internet use,

 including 71% of blacks and 59% of all Latinos, both English-dominant and Spanish-dominant.

 Cell phone access has also been associated with knowledge of health information.

This article describes the results of a pilot study conducted to evaluate the feasibility of utilizing low-cost, mobile HIT to support chronic disease self management between clinic visits among patients in an urban safety net setting. Cell phone text messaging was used to provide blood sugar measurement prompts and appointment reminders to adult patients with diabetes in order to promote engagement in self-care behaviors.

The quasi-experimental pilot study was conducted among English-speaking and Spanish-speaking adult patients with diabetes who have access to cell phones and who regularly receive primary healthcare at Sam Sandos Westside Family Health Center (Westside), 1 of 8 federally qualified community health centers in the Denver Health integrated healthcare system in Denver, Colorado. Westside serves a predominantly Latino population (81%) that is largely either uninsured (41%) or on Medicaid or Medicare (56%).

The Colorado Multiple Institutional Review Board approved this study prior to enrollment and implementation activities. All English-dominant and Spanish-dominant adult patients included in the Denver Health diabetes registry were identified as potential study participants. Westside providers and clinic staff confirmed potential participants’ access to cell phones at clinic visits and referred patients to the study nurse for recruitment. All patients signed a written consent prior to participation.

A software platform, the Patient Relationship Manager (PRM), was created in partnership between Denver Health, EMC Consulting, and Microsoft Corporation to handle text messaging activities. The PRM system was designed to automatically send text messages to patients according to an established schedule, and to process text message responses received from patients for appropriate action based on established threshold values.

Participants (N = 47) received text message prompts in their choice of English or Spanish over a 3-month period. Messages requesting fasting blood sugar readings were sent to each patient at 7:15 am on Monday, Wednesday, and Friday of each week during the study period, and appointment reminder messages were sent to each patient 3 times in advance of each scheduled appointment that patient had during the study period, at 7 days, 3 days, and 1 day prior to the appointment date. Blood sugar requests included the day of the week and calendar date that the message was sent and a request for the patient to respond with his or her fasting blood sugar that day in numeric values. Appointment reminders included the time, date, and location of the upcoming appointment along with a request for patients to respond with “Y” or “N” to indicate whether or not they planned to attend the appointment. The system automatically returned an acknowledgment of receipt in response to all patient-sent messages. Examples of all message types are shown in 

Misformatted responses and fasting blood sugar values reported by patients as either above or below an established clinical range of 70 to 400 mg/dL were automatically flagged by PRM and routed to a work queue for further action. A registered nurse dedicated to the study at 0.2 full-time equivalents reviewed all flagged messages and contacted patients by telephone for follow-up assessment. The study nurse problem-solved message formatting issues with patients and presented out-of-range values to a physician champion, who helped formulate a targeted response plan to improve the patient’s glucose control (eg, by changing eating patterns or adjusting medications). The nurse also ensured that both telephone encounters and patient-reported blood sugar measurements were appropriately documented in the medical record.

Outcome measures included appointment attendance rates and patient response rates to text message prompts. A power analysis was performed using Power Analysis and Sample Size (PASS 2008) software (NCSS, Kaysville, Utah), with the significance level (alpha) targeted at .05 with greater than 90% certainty to identify the number of appointments needed to detect a significant absolute difference of 10% in preintervention and postintervention attendance rates. Clinical chart review for all study patients was undertaken for the last 2 clinic visits prior to study enrollment to assess how often patients discussed home glucose readings with their providers.

Focus groups were conducted with patients in order to assess satisfaction with the HIT intervention program by soliciting feedback on the experience and recommendations for future improvement. Two focus groups were held, 1 conducted in English and 1 in Spanish with the aid of an experienced medical interpreter. Both group sessions were audio recorded. Recordings were supplemented with written notes taken by the group facilitator. Focus group data were obtained through inductive analysis of the audio recordings and written notes. Emergent themes and patterns among patients’ remarks were identified and synthesized into a summary report.

Demographics of both study participants and the overall population of Westside patients with diabetes are presented in 

. Patients of all ages were active participants in the program,with the majority of participants aged 40 to 49 years (n = 14), 50 to 59 years (n = 17), and 60 to 69 years (n = 9). The intervention group was significantly younger than the clinic population, but well matched for sex and racial/ethnic background.

A total of 1585 text messages requesting a response were automatically sent to patients by the PRM system. Patients responded to prompts with a total of 1080 text messages (68.14%), of which almost all were correctly formatted (1066 messages; 98.7%). More than three-fourths (79%) of the cohort responded regularly to more than 50% of their message prompts. Home glucometer result availability was significantly improved among study participants. Two-thirds (66.4%) of study patients provided correctly formatted blood sugar values in response to medical measurement text message requests received during the study period. By contrast, chart review data for each study participant indicated that providers were able to review home glucometer readings with patients only 12% of the time during the last 2 clinic visits prior to the intervention.

Despite reminders for 69 appointments during the intervention, cancellation rates and no-show rates were 21% and 14%, respectively, and did not differ from preintervention rates. However, the power analysis indicated that a sample size of 349 appointments was needed to achieve adequate power to detect a significant difference in appointment attendance rates.

A total of 8 patients (6 female and 2 male) participated in 2 focus groups. Among these participants, 2 were aged 30 to 39 years, 2 were aged 50 to 59 years, and 4 were aged 60 to 69 years. Two participants were Spanish dominant and the remaining 6 participants were English dominant. Half of the focus group participants were white, and 37.5% were Latino. Participants’ comfort level with program technology, perception of and satisfaction with the intervention program (including message type and frequency), and the perceived impact of the program on diabetes self management were explored as major domains in group discussions.

All 8 participants reported comfort with the process of text messaging, both in sending and receiving messages. Three participants identified text messaging as a means they already used for communicating with their children. One participant with vision impairment involved other family members to assist with reading and replying to messages. Three participants expressed a preference for text messaging over other computer-based communications such as e-mail.

Participants’ Satisfaction With the Program

All 8 participants viewed the text message program favorably, expressing both a liking for it and a desire for it to continue, along with a willingness to recommend it to others. The program was perceived as having expanded participants’ communication and engagement with the healthcare system; half of the participants remarked on feeling that their providers were “more connected” or otherwise aware of what was going on with them. Financial costs associated with text messaging were perceived by participants to be negligible; all participants reported having unlimited text messaging plans. Five participants indicated that they would be willing to pay for program messages even if they did not have unlimited plans.

All participants found the frequency of text messaging during the study period to be acceptable. Half of the participants expressed interest in receiving daily messages, while the other half were content with receiving messages according to the current schedule. No one wanted to receive messages less often. Four participants expressed specific appreciation for being able to respond to messages throughout the day on which they were requested, according to the patient’s own time frame and personal schedule. Six participants noted that they also liked the acknowledgment from the system that their responses had been received. Although all participants received self-management support messages, not all received appointment reminders. Two who did expressed appreciation for them, and the others indicated interest in receiving them.

Program Impact on Diabetes Self Management

In addition to a general overall perception of usefulness, the program was specifically considered to be helpful in establishing regular morning routines (4 participants) and in improving recollection of and adherence to self-management activities (4 participants). Participants reported increased awareness of their health information (2 participants) and feelings of being made accountable for their choices (4 participants), along with better perceived social support either from family (2 participants) or from the system itself (3 participants) in the absence of other established social groups. Three participants observed that they would like additional, personalized feedback about their performance on self-management activities.

In this quasi-experimental pilot study, we assessed the feasibility of engaging both English-speaking and Spanishspeaking adults with diabetes in self-management behaviors between clinic visits by using cell phone text messaging to provide blood sugar measurement prompts and appointment reminders. Over the course of the study, there was a sustained high response rate of 68% to all text message prompts received, with 66% of patients providing home glucose data upon request and 79% of patients responding to more than half of their glucose prompts. Focus group participants categorically embraced the text message—based outreach.

This pilot intervention differed from usual care by engaging patients between clinic visits. Like many safety net health centers, Denver Health does not have the capability to support downloading home glucose reading data into the clinical system during clinic visits. Depending solely on clinic visits to support discussion of self-monitoring data is challenging as well: chart review of 2 clinic visits prior to the intervention for each study participant indicated that providers reviewed home glucometer readings only 12% of the time. Although in its infancy, text messaging has been used to improve the management of asthma, hypertension, and diabetes,

 but we are not aware of studies conducted in an indigent, predominantly Latino population.

The results of this pilot study demonstrate that there is potential for using cell phone text messaging to support chronic disease management and engagement in an underserved population. Additionally, the sustained response rate and consistent response patterns demonstrated among study participants enhances the ability of the patient and provider to engage in shared decision making at clinic visits by improving access to self-monitoring glucose data collected in sufficient volume to guide well-informed clinical decisions.

Cell phoneâ€œbased text messaging may be used to feasibly support chronic disease management and engagement in diabetes self-care behaviors for some patients.

Care by Cell Phone: Text Messaging for Chronic Disease Management

To analyze the impact of worksite wellness programs on health and financial outcomes, and the effect of incentives on participation.

Sources were PubMed, CINAHL & EconLit, Embase, Web of Science, and Cochrane for 2000-2011. We examined articles with comparison groups that assessed health-related behaviors, physiologic markers, healthcare cost, and absenteeism. Data on intervention, outcome, size, industry, research design, and incentive use were extracted.

 A total of 33 studies evaluated 63 outcomes. Positive effects were found for threefourths of observational designs compared with half of outcomes in randomized controlled trials. A total of 8 of 13 studies found improvements in physical activity, 6 of 12 in diet, 6 of 12 in body mass index/weight, and 3 of 4 in mental health. A total of 6 of 7 studies on tobacco and 2 of 3 on alcohol use found significant reductions. All 4 studies on absenteeism and 7 of 8 on healthcare costs estimated significant decreases. Only 2 of 23 studies evaluated the impact of incentives and found positive health outcomes and decreased costs.

The studies yielded mixed results regarding impact of wellness programs on healthrelated behaviors, substance use, physiologic markers, and cost, while the evidence for effects on absenteeism and mental health is insufficient. The validity of those findings is reduced by the lack of rigorous evaluation designs. Further, the body of publications is in stark contrast to the widespread use of such programs, and research on the effect of incentives is lacking.

Because of the emphasis the Affordable Care Act places on worksite health promotion, employers have increasingly offered worksite wellness programs, but little research has evaluated the current impact of these programs.

		We analyzed a total of 33 studies published since 2000 that evaluated the characteristics, impact, and incentives of worksite wellness programs.

		 Studies suggest mixed impact on health-related behaviors and cost, with insufficient evidence for effects on absenteeism and mental health.

		 Lack of rigorous evaluation designs reduces the internal validity of these findings.

Employers have increasingly offered worksite wellness programs to employees and their families to decrease their cost of providing healthcare coverage and improve their employees’ productivity. The goals of these programs are to promote healthy lifestyles and prevent disease with educational (eg, diet counseling) and motivational (eg, provision of incentives for lifestyle changes) approaches.

 In 2009, 58% of US employers offered at least 1 wellness program.

 In 2010, consumer participation in programs rose from 19% to 22%.

 This trend is likely to accelerate, as the Patient Protection and Affordable Care Act emphasizes prevention.

 The law provides wellness program start-up grants for small firms, establishes a 10-state demonstration program to reward program participation, and establishes technical assistance for evaluating programs. The law also gives employers greater latitude in rewarding staff for healthy lifestyles by raising the rewards for program participation. The limit, set at 20% of the cost of coverage, will increase to 30% in 2014, and the secretaries of Health and Human Services, Labor, and the Treasury will jointly have the authority to raise it as high as 50%.

Reflecting their growing importance, several reviews of worksite wellness programs and their components have been published. Baicker et al

assessed the impact of 32 programs on medical costs and absenteeism that were published since 1982 and found that programs typically return 3 dollars for every dollar invested, which is consistent with other research suggestive of savings.

Other studies have found positive effects as well,

 including evidence for certain components of wellness programs (eg, health risk assessment).

Despite these findings, a review is needed on the current impact of wellness programs to examine how employers have responded to current policy and programmatic priority changes. We conducted a systematic review of wellness programs to understand:

• What are the characteristics of the worksite wellness

• What impact do programs have on outcomes?

• What types of incentives are provided for program uptake and what is their impact?

This review contributes to the evidence base in several ways. We considered outcomes beyond medical cost and absenteeism to include behavior change and health effects. Only studies with a comparison strategy were included, to maximize validity of the findings. We also limited our scope to studies about comprehensive programs that were published after 2000 to yield a more accurate reflection of current programs. Finally, we looked at the use of incentives for program uptake.

We conducted a keyword search covering PubMed, CINAHL & EconLit (EBSCO), Embase, Web of Science, and Cochrane from January 2000 through June 2011 (see the 

). Additional articles were identified through reference searches of recent literature reviews or meta-analyses.

Articles were included if they had a control or other comparison group and evaluated outcomes of comprehensive worksite wellness programs (ie, multiple wellness components focused on health promotion or disease prevention). We excluded opinion and theory articles, reviews, articles without a comparison group, non-English language and non-US articles, articles published before 2000, and articles that focused exclusively on disease management.

Two investigators (KCO, KVB) independently evaluated articles for inclusion based on title and abstract review, and then full text review. A third investigator (SM or CS) served as tie breaker in case of discrepancy.

We extracted type of intervention, setting, and research design from each study. Programs and worksites were classified by type, size, and industry.

We categorized the quality of the design using methods adapted from previous meta-analyses

: controlled trials with random assignment, prospective studies with nonrandomly assigned comparison groups, and observational designs with internal comparison groups (eg, participants vs nonparticipants).

We identified 1546 articles from our search and 9 through bibliography searches of review articles (

). We excluded 1492 upon title and abstract review. A total of 62 full-text articles were assessed for eligibility; 29 were not eligible (eg, noncomprehensive programs, international, no comparison), yielding a final sample of 33 articles.

Of the 33 studies, 22 reported company size (

); 8 studies were done in medium-sized companies, while only 1 reported on a small worksite. Of the 33 studies, 29 reported the industry. About half were conducted in companies that provided services, which is comparable to the distribution of industries in the overall economy.

Program modality varied substantially, as 31 studies reported multiple delivery methods (23 reported 3 or more). The most common modality was self-help or educational materials and/or individual coaching or counseling.

A total of 63 outcomes were evaluated between the 33 studies (

). The most common were exercise (n = 13), diet (n = 12), and physiologic markers (n = 12). Others reported on healthcare cost (n = 8), smoking (n = 7), alcohol use (n = 3), absenteeism (n = 4), and mental health (n = 4). The majority of studies (64%) used self-reported data for at least 1 outcome. About three-fourths of the observational designs reported beneficial outcomes compared with about half of the randomized trials.

Analysis of studies of worksite wellness programs suggested mixed impact on health-related behaviors and cost, with insufficient evidence regarding absenteeism and mental health.

Systematic Review of the Impact of Worksite Wellness Programs

 The Patient Assessment of Chronic Illness Care (PACIC) is a widely used instrument to evaluate the quality and patient-centeredness of chronic illness care based on the Chronic Care Model (CCM). It is a validated and reliable instrument which consists of 20 items. Additionally, a short form with 11 items was developed. The aim of this study was to translate this short form into German and examine the psychometric properties among patients with a chronic illness in Germany.

We performed a translation and cultural adaptation of the PACIC short form into German. The German version was externally validated with the 20-item PACIC. Cronbach a, descriptive statistics, and principal component analysis were used to assess psychometric properties.

 In total, 264 primary care patients completed the PACIC short form. The PACIC short form showed good convergent construct validity to the 20-item PACIC (Spearman rank correlation 0.82, 

 <.001) and high internal consistency (Cronbach a 0.87). Principal component analysis underlined the 1-dimensional structure of the instrument. No correlation between the mean overall score of the PACIC short form and the number of chronic conditions (r = 0.068; 

 The PACIC short form showed good to very good psychometric properties and reliable measures regarding patient assessment of receiving care congruent with the CCM. It is a less burdensome instrument which can be used for further research of patients with more than 1 chronic condition.

)This study looked at the psychometric properties of the German version of the Patient Assessment of Chronic Illness Care (PACIC) short form among patients with chronic illness.

		A major challenge for future healthcare is the care of patients with chronic conditions.

		Results showed that nearly 50% of respondents had more than 3 chronic conditions.

		The PACIC short form is a reliable instrument with good psychometric properties and can be used for patients with more than 1 chronic condition.

Care for patients with chronic conditions is one of the major challenges in future healthcare.

 Patients with a chronic condition seek medical care more often than patients without a chronic condition. This may increase the cost of healthcare.

 The Chronic Care Model (CCM), developed by Wagner et al, illustrates a multidimensional framework for improving the quality of chronic care based on 6 key dimensions: organization of healthcare, clinical information system, delivery system design, decision support, selfmanagement support, and community resources.

The Patient Assessment of Chronic Illness Care (PACIC) is an instrument widely used to evaluate the quality and patient-centeredness of chronic illness care.

 It has been developed to measure implementation of the CCM at the level of patients assessing the behavior of professionals and practice teams within their care.

 It is a validated and reliable instrument for different chronic conditions

 and has been translated into different languages.

However, it was found that the development of the structure of the PACIC instrument implicates some methodological limitations.

 The 20-item PACIC based on a 5-factor solution (patient activation, delivery system design/decision support, goal setting/tailoring, problem solving/contextual, and follow-up/coordination) was evaluated with a confirmatory factor analysis. Some research groups showed that this determination of factorial validity, including the distribution of items as well as the option of handling the responses as interval or ratio levels, is the main point of criticism.

Therefore, a PACIC short form with an adapted response scale was developed to evaluate patient assessment of receiving care congruent with the CCM.

The aim of this study was to translate and examine the psychometric properties of the PACIC short form among patients in Germany with a chronic illness.

We performed an observational study in 11 general practices in Germany (8 located in the federal state of Baden-Württemberg, and 3 in the federal\ state of Thuringia). All practices were teaching sites for medical students of either the University Hospital of Heidelberg or the University Hospital of Jena.

To adapt the PACIC short form we followed the Principles of Good Practice for the Translation and Cultural Adaptation Process by the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) task force

 as follows: We received permission from the authors who developed the PACIC short form, Gugiu et al from Western Michigan University, to translate and adapt a German version of the instrument.

 Two researchers (JS, TF) independently translated the English version of the PACIC short form 11-item scale into German. Divergent results were discussed during consensus meetings with a third researcher (KG). After a linguistic adaptation, no item was assumed to be completely inappropriate.

For recruiting practices we chose a convenience sample of teaching practices working either with the University Hospital of Heidelberg or the University Hospital of Jena. Within the practices of the participating general practitioners (GPs), patients 18 years or older, suffering from at least 1 major chronic condition (defined in accordance with the German Social Code Book V §62),

 were asked to participate in the survey. Patients with severe cognitive impairment or significant language barriers were excluded from the study by the individual GP of the practice. Patients were asked to fill out a depersonalized paper-based questionnaire and to send it back to the study center. We provided a postage-free envelope but no further financial incentives for patients. Written informed consent was obtained from each participant. At each practice site in Baden-Württemberg, the questionnaire was given to 50 participating patients (45 patients in each of the 3 practice sites in Thuringia). The depersonalized paper-based questionnaire consisted of sociodemographic data, a list of 20 chronic conditions, the 20-item PACIC, and the PACIC short form. All participating practices were supported for recruiting patients with 50 €.

The PACIC short form consists of 11 items with an 11-point percentage scale ranging from 0% (“Never”) to 100% (“Almost”). In addition to the PACIC short form, we measured sociodemographic data with a set of questions from a German standard questionnaire and patients were also asked to select their chronic conditions from a list of 20 conditions, which were used in previous evaluation studies on chronic illness care.

 The sociodemographic data included questions regarding age, gender, marital status, and education (

). We used the 20-item PACIC to assess convergent construct validity with the PACIC short form, which has been validated within multiple studies.

 The items of the long version were scored on a 5-point Likert scale, ranging from “1” (“no/never”) to full accordance “5” (“yes/always”).

All analyses were carried out using SPSS 18.0 software (SPSS Inc, Chicago, Illinois). The aim of the study was to assess psychometric properties of the PACIC short form and to examine the convergent construct validity with the 20-item PACIC. The reliability was assessed by using Cronbach’s alpha, which indicates whether an item of a scale is appropriate for assessing the underlying concept of its scale.

 Values for Cronbach’s alpha range from 0 to 1; the closer they are to 0 the less the items are related to one another. Values above 0.60 are generally considered to indicate satisfying internal consistency and values above 0.80 indicate a high internal consistency. We performed principal factor analysis (eigenvalue >1, varimax rotation) and determined the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and the Bartlett’s test of sphericity. Convergent construct validity was assessed in terms of a Spearman rank correlation test between means of each item of the PACIC short

form and the mean for the 20-item PACIC overall score, a  well as between the mean score of the PACIC short form and the mean score of the 20-item PACIC. In this context, correlations often range between 0.2 and 0.6, rarely above; correlations between 0.40 and 0.60 are regarded as good correlations.

 Frequency distributions and statistical moments were calculated as percentages of patients rating at each level of item. We determined means, standard deviations of means, and missing values on item level to identify potentially inadequate items. Finally, we examined with Spearman rank correlation test the relation between the mean overall score of the PACIC short form and the number of chronic conditions. An alpha level of P 

.05 was used for tests of statistical significance.

The study was fully approved by the ethics committees of the medical faculties of the University of Heidelberg and the University of Jena.

In total, 264 out of 535 participants (49%) completed the PACIC short form. Details on sociodemographic characteristics and morbidity are given in Table 1. Almost every second participant (47.0%) has more than 3 chronic conditions. Hypertension (n = 184; 69.7%), back pain (n = 148; 56.1%), osteoarthritis (n = 113; 42.8%), and type 2 diabetes (n = 86; 32.6%) were the most frequently reported chronic conditions.

Factor analysis revealed a 1-dimensional structure of the PACIC short form with explained variance of R² = 48.15% (KMO 0.90, Barlett's test of sphericity 

 <.001). The factor loadings ranged between 0.52 and 0.85. The Spearman rank correlation coefficients ranged between 0.38 for item 6 of the PACIC short form and 0.78 for item 8 of the PACIC short form. For the mean overall scores of the short version and 20- item PACIC, the correlation was 0.82 (

 <.001). Details on factor loadings and Spearman rank correlation are given in 

. Furthermore, the PACIC short form showed an internal consistency reliability of 0.87 (Cronbach α).

Frequency distributions and statistical moments of the PACIC short form are presented in 

. The participants tended to gravitate to both end points, 0% and 100%. The skewness of most of the items was within tolerable levels, tending to be fairly close to zero. A majority of the kurtosis values removed substantially from zero, which is acceptable.

 displays details on missing values on item level. We observed non-response rates which range from 4.2% and 12.5% on item level. The highest score of missing values showed item 7 of the PACIC short form with 33 nonresponders (12.5%). The lowest score was shown among item 2 of PACIC short form with 11 non-responders (4.2%).

No correlation between the mean overall score of the PACIC short form and the number of chronic conditions (r = 0.068; 

The presented study describes good or even very good psychometric properties of the German version of the PACIC short form and shows reliable measures regarding patient assessment of receiving care congruent with the CCM. The results of our study found high internal consistency and good external validity of the 1-dimensional scale. Moderate missing rates, low floor, and ceiling effects on item level support these results.

Compared with the original version of the PACIC short form, the German version shows similar psychometric properties. The development of the original instrument based on data from 2 samples of patients with type 2 diabetes revealed an internal consistency of α = 0.95 and α = 0.96, respectively.

The high correlation between the mean scores of both PACIC versions shows that the short form of the PACIC reflects the 20-item PACIC sufficiently. Moreover, regarding the fact that the mean score of the PACIC short form and the number of chronic conditions do not correlate, we can conclude that the PACIC measurement is independent of the number of chronic conditions.

 The development of a generic questionnaire to assess whether quality of care is in focus with the CCM is important for future studies focusing on the issue of multimorbidity.

 The number and distribution of chronic conditions which we observed are comparable with other studies.

Three items (“given choices about treatment to think about” [PACIC 1 S]; “helped to make a treatment plan that I could carry out in my daily life” [PACIC 7 S]; and “helped to plan ahead so I could take care of my condition even in hard times” [PACIC 8 S]) had about 10% missing values. We could assume that these 3 items are not explicit enough, as there are 2 issues to assess in each question, and therefore they need reformulation.

Two items, “choices about treatment” and “treatment plan,” resulted in a relatively low response rate. From shared decision-making (SDM) studies in Germany we know that patient-doctor communication is an important subject for the individual patient’s treatment preference.

For future studies focusing on quality of care from the background of the Chronic Care Model, the PACIC short form is appropriate.

February 2012

Identifying Patients With Osteoporosis or at Risk for Osteoporotic Fractures

Care by Cell Phone: Text Messaging for Chronic Disease Management

Systematic Review of the Impact of Worksite Wellness Programs

Adaptation and Psychometric Properties of the PACIC Short Form

EHRs in Primary Care Practices: Benefits, Challenges, and Successful Strategies

Nurse-Run, Telephone-Based Outreach to Improve Lipids in People With Diabetes

A Dementia Care Management Intervention: Which Components Improve Quality?

Early Evaluations of the Medical Home: Building on a Promising Start

Hospital Readmission Rates in Medicare Advantage Plans

TO THE EDITORS: