Trends and predictors of readmission after catheter ablation for atrial fibrillation, 2009-2013.

UNLABELLED: As the number of patients undergoing catheter ablation for atrial fibrillation (AF) increases, there is a growing focus on optimizing the quality and efficiency of. Readmission is often considered an indicator of both quality and efficiency of care delivery. We sought to estimate rates and identify predictors of readmission after catheter ablation. METHODS AND RESULTS: Using a large, national administrative claims database, we identified all AF patients who underwent catheter ablation between January 2009 and December 2013 (10,705 ablation cases). We examined incident readmission and the primary diagnosis during the readmission episode of care. We used Cox proportional hazard models to identify associations between readmission and patient and institutional characteristics. A total of 1,433 (13.4%) ablation patients were readmitted within 90 days of ablation for any cause, and 573 (5.4%) were admitted with AF as the primary diagnosis. There was a decline in all-cause (from 15.6% to 12.8%; P = .04) and AF-related (6.4%-5.0 %; P = .03) 90-day readmission over the study period. In a multivariate model, earlier year of ablation and each of 9 chronic conditions (alone or in combination) were independently associated with risk of readmission. CONCLUSIONS: Between 2009 and 2013, there was a reduction in 90-day readmission rates after AF ablation, suggesting improved periprocedural care of these patients. Identifying patients at high risk for readmission after catheter ablation for AF may offer an opportunity for early intervention and, ultimately, reduction in procedural morbidity and medical costs.

Use of Post-acute Care Services and Readmissions After Left Ventricular Assist Device Implantation in Privately Insured Patients.

BACKGROUND: Very little is known about health care resource utilization, including post-acute care use and hospital readmissions, after left ventricular assist device (LVAD) implantation. METHODS AND RESULTS: Administrative claims from a database of multiple United States health plans were used to identify patients that received an LVAD (ICD-9 code 37.66) and survived to hospital discharge from January 1-2006, through September 30-2013. Post-acute care use was defined as a skilled nursing facility or rehabilitation stay within 90 days after hospital discharge. Patients were censored at heart transplantation or end of coverage through December 31-2013. Of 583 patients (mean age 55 years, 77% male), 223 (38.3%) used post-acute care services, more commonly in patients with diabetes, who required hemodialysis, and who had LVADs implanted at hospitals in more populated areas, with more beds, and in the northeast region (P < .05 for each). The most common reasons for readmission were device complications, heart failure, and arrhythmia. Readmission risk was higher in patients who had diabetes, peripheral vascular disease, and longer hospital length of stay, but it did not differ by post-acute care use. CONCLUSIONS: Use of post-acute care services varies based on hospital characteristics. We found no association between post-acute care use and readmission risk after LVAD implantation.

Comparative risk of gastrointestinal bleeding with dabigatran, rivaroxaban, and warfarin: population based cohort study.

OBJECTIVE: To determine the real world risk of gastrointestinal bleeding associated with the use of the novel oral anticoagulants dabigatran and rivaroxaban compared with warfarin. DESIGN: Retrospective, propensity matched cohort study. SETTING: Optum Labs Data Warehouse, a large database including administrative claims data on privately insured and Medicare Advantage enrollees. PARTICIPANTS: New users of dabigatran, rivaroxaban, and warfarin from 1 November 2010 to 30 September 2013. MAIN OUTCOME MEASURES: Incidence rates (events/100 patient years) and propensity score matched Cox proportional hazards models were used to estimate rates of total gastrointestinal bleeds, upper gastrointestinal bleeds, and lower gastrointestinal bleeds for the novel oral anticoagulants compared with warfarin in patients with and without atrial fibrillation. Heterogeneity of treatment effect related to age was examined using a marginal effects model. RESULTS: The incidence of gastrointestinal bleeding associated with dabigatran was 2.29 (95% confidence interval 1.88 to 2.79) per 100 patient years and that associated with warfarin was 2.87 (2.41 to 3.41) per 100 patient years in patients with atrial fibrillation. In non-atrial fibrillation patients, the incidence of gastrointestinal bleeding was 4.10 (2.47 to 6.80) per 100 patient years with dabigatran and 3.71 (2.16 to 6.40) per 100 patient years with warfarin. With rivaroxaban, 2.84 (2.30 to 3.52) gastrointestinal bleeding events per 100 patient years occurred in atrial fibrillation patients (warfarin 3.06 (2.49 to 3.77)/100 patient years) and 1.66 (1.23 to 2.24) per 100 patient years in non-atrial fibrillation patients (warfarin 1.57 (1.25 to 1.99)/100 patient years). In propensity score matched models, the risk of gastrointestinal bleeding with novel oral anticoagulants was similar to that with warfarin in atrial fibrillation patients (dabigatran v warfarin, hazard ratio 0.79 (0.61 to 1.03); rivaroxaban v warfarin, 0.93 (0.69 to 1.25)) and in non-AF patients (dabigatran v warfarin, hazard ratio 1.14 (0.54 to 2.39); rivaroxaban v warfarin, 0.89 (0.60 to 1.32)). The risk of gastrointestinal bleeding increased after age 65, such that by age 76 the risk exceeded that with warfarin among atrial fibrillation patients taking dabigatran (hazard ratio 2.49 (1.61 to 3.83)) and patients with and without atrial fibrillation taking rivaroxaban (2.91 (1.65 to 4.81) and 4.58 (2.40 to 8.72), respectively). CONCLUSIONS: The risk of gastrointestinal bleeding related to novel oral anticoagulants was similar to that for warfarin. Caution should be used when prescribing novel oral anticoagulants to older people, particularly those over 75 years of age.

Risk of stroke after catheter ablation versus cardioversion for atrial fibrillation: A propensity-matched study of 24,244 patients.

BACKGROUND: Stroke is the major cause of morbidity and mortality related to atrial fibrillation (AF). Catheter ablation for AF is effective in reducing AF burden, but its impact on long-term stroke risk is unknown. OBJECTIVE: We sought to evaluate the periprocedural and long-term stroke risk after catheter ablation or cardioversion for AF. METHODS: This retrospective, propensity-matched study using a national administrative claims database identified patients with AF who underwent catheter ablation and a comparison group (matched on age, sex, year of treatment, CHA2DS2-Vasc score, and Charlson index) who underwent cardioversion between 2005 and 2012. The primary end points were (1) time to first ischemic or hemorrhagic stroke or transient ischemic attack (TIA) and (2) time to first ischemic or hemorrhagic stroke excluding TIA. We compared periprocedural incident stroke (within 30 days of ablation or cardioversion) as well as total strokes between the 2 groups. RESULTS: A total of 24,244 patients (12,122 patients undergoing ablation and 12,122 patients undergoing cardioversion) were included in the analysis. Incident periprocedural stroke or TIA occurred in 0.5% of the ablation group and 0.3% of the cardioversion group (P = .04). There was a significant initial risk of stroke/TIA with ablation within the first 30 days (rate ratio 1.53; P = .05). After 30 days, this risk was significantly lower in the ablation group (rate ratio 0.78; P = .03). CONCLUSION: In patients with AF, there is a small periprocedural stroke risk with ablation in comparison to cardioversion. However, over longer-term follow-up, ablation is associated with a slightly lower rate of stroke.

Trends in computed tomography utilization rates: a longitudinal practice-based study.

OBJECTIVES: Computed tomography (CT) use has increased dramatically over the past 2 decades, leading to increased radiation exposure at the population level. We assessed trends in CT use in a primary care (PC) population from 2000 to 2010. METHODS: Trends in CT use from 2000 to 2010 were assessed in an integrated, multi-specialty group practice. Administrative data were used to identify patients associated with a specific primary care provider and all CT imaging procedures. Utilization rates per 1000 patients and CT rates by type and medical specialty were calculated. RESULTS: Of 179,032 PC patients, 55,683 (31%) underwent CT. Mean age (SD) was 31.0 (23.6) years; 53% were female patients. In 2000, 178.5 CT scans per 1000 PC patients were performed, increasing to 195.9 in 2010 (10% absolute increase, P = 0.01). Although utilization rates across the 10-year period remained stable, emergency department (ED) CT examinations rose from 41.1 per 1000 in 2000 to 74.4 per 1000 in 2010 (81% absolute increase, P < 0.01). CT abdomen accounted for more than 50% of all CTs performed, followed by CT other (19%; included scans of the spine, extremities, neck and sinuses), CT chest (16%), and CT head (14%). Top diagnostic CT categories among those undergoing CT were abdominal pain, lower respiratory disease, and headache. CONCLUSIONS: Although utilization rates across the 10-year period remained stable, CT use in the ED substantially increased. CT abdomen and CT chest were the two most common studies performed and are potential targets for interventions to improve the appropriateness of CT use.

Perceptions and practices of therapeutic hypothermia in American neonatal intensive care units.

OBJECTIVE: In 2005, therapeutic hypothermia (TH) was used in few American neonatal intensive care units (NICUs) with great variability in practices. We hypothesized that TH would be used with greater frequency and uniformity today. STUDY DESIGN: We surveyed directors of 797 NICUs queried in our prior study to determine attitudes toward and practices of TH. RESULTS: Of the 781 participants with valid addresses, we received completed surveys from 330 (42.3%). There was an increase in the number of respondents who believed that TH is effective (85% versus 31%, p < 0.0001). More NICUs used TH (50% versus 6%, p < 0.0001) and nearly all not offering TH transferred eligible neonates to centers that did (97% versus 29%, p < 0.0001). There has been increased standardization of TH practices with regard to enrollment criteria, duration, and methods of monitoring. CONCLUSION: TH has become standard of care for the treatment of HIE in the United States. Most NICUs that use TH adhere to protocols, but variation still exists in TH practices.

Risk-stratification methods for identifying patients for care coordination.

BACKGROUND: Care coordination is a key component of the patient-centered medical home. However, the mechanism for identifying primary care patients who may benefit the most from this model of care is unclear. OBJECTIVES: To evaluate the performance of several risk-adjustment/stratification instruments in predicting healthcare utilization. STUDY DESIGN: Retrospective cohort analysis. METHODS: All adults empaneled in 2009 and 2010 (n = 83,187) in a primary care practice were studied. We evaluated 6 models: Adjusted Clinical Groups (ACGs), Hierarchical Condition Categories (HCCs), Elder Risk Assessment, Chronic Comorbidity Count, Charlson Comorbidity Index, and Minnesota Health Care Home Tiering. A seventh model combining Minnesota Tiering with ERA score was also assessed. Logistic regression models using demographic characteristics and diagnoses from 2009 were used to predict healthcare utilization and costs for 2010 with binary outcomes (emergency department [ED] visits, hospitalizations, 30-day readmissions, and highcost users in the top 10%), using the C statistic and goodness of fit among the top decile. RESULTS: The ACG model outperformed the others in predicting hospitalizations with a C statistic range of 0.67 (CMS-HCC) to 0.73. In predicting ED visits, the C statistic ranged from 0.58 (CMSHCC) to 0.67 (ACG). When predicting the top 10% highest cost users, the performance of the ACG model was good (area under the curve = 0.81) and superior to the others. CONCLUSIONS: Although ACG models generally performed better in predicting utilization, use of any of these models will help practices implement care coordination more efficiently.

Hospitalizations and emergency department use in Mayo Clinic Biobank participants within the employee and community health medical home.

OBJECTIVE: To evaluate the participants in the Mayo Clinic Biobank for their representativeness to the entire Employee and Community Health program (ECH) primary care population with regard to hospital utilization. PATIENTS AND METHODS: Participants enrolled in the Mayo Clinic Biobank from April 1, 2009, to December 31, 2010, were linked to the ECH population. These individuals were categorized into risk tiers (0-4) on the basis of the number of health conditions present as of December 31, 2010. Outcomes were ascertained through December 31, 2011. Hazard ratios (HRs) and 95% CIs for risk of hospitalization, emergency department (ED) visits, and for risk of hospitalization and emergency department (ED) visits were estimated. RESULTS: The 8927 Biobank participants were part of ECH (N=84,872). Compared with the entire ECH population, the Biobank-ECH participants were more likely to be female (64.3% vs 54.6%), older (median age, 58 years vs 47 years), and categorized to tier 0 (6.4% vs 24.0%). There were strong positive associations between tier (tier 4 vs combined tiers 0 and 1) and risk of hospitalization (HR, 5.8; 95% CI, 4.6-7.5) and ED visits (HR, 5.4; 95% CI, 4.2-6.8) among Biobank-ECH participants. Similar associations for risk of hospitalization (HR, 8.5; 95% CI, 7.8-9.3) and ED visits (HR, 6.9; 95% CI, 6.4-7.5) were observed for the entire ECH population. CONCLUSION: Although the Biobank-ECH participants were older and had more chronic conditions compared with the overall ECH population, the associations of risk tier with utilization outcomes were similar, supporting the use of the Biobank participants to assess biomarkers for health care outcomes in the primary care setting.

Enhanced recovery in gynecologic surgery.

OBJECTIVE: To investigate the effects of enhanced recovery (a multimodal perioperative care enhancement protocol) in patients undergoing gynecologic surgery. METHODS: Consecutive patients managed under an enhanced recovery pathway and undergoing cytoreduction, surgical staging, or pelvic organ prolapse surgery between June 20, 2011, and December 20, 2011, were compared with consecutive historical controls (March to December 2010) matched by procedure. Wilcoxon rank-sum, χ, and Fisher's exact tests were used for comparisons. Direct medical costs incurred in the first 30 days were obtained from the Olmsted County Healthcare Expenditure and Utilization Database and standardized to 2011 Medicare dollars. RESULTS: A total of 241 enhanced recovery women in the case group (81 cytoreduction, 84 staging, and 76 vaginal surgery) were compared with women in the control groups. In the cytoreductive group, patient-controlled anesthesia use decreased from 98.7% to 33.3% and overall opioid use decreased by 80% in the first 48 hours with no change in pain scores. Enhanced recovery resulted in a 4-day reduction in hospital stay with stable readmission rates (25.9% of women in the case group compared with 17.9% of women in the control group) and 30-day cost savings of more than $7,600 per patient (18.8% reduction). No differences were observed in rate (63% compared with 71.8%) or severity of postoperative complications (grade 3 or more: 21% compared with 20.5%). Similar, albeit less dramatic, improvements were observed in the other two cohorts. Ninety-five percent of patients rated satisfaction with perioperative care as excellent or very good. CONCLUSIONS: Implementation of enhanced recovery was associated with acceptable pain management with reduced opioids, reduced length of stay with stable readmission and morbidity rates, good patient satisfaction, and substantial cost reductions. LEVEL OF EVIDENCE: II.

30-day hospital readmission of older adults using care transitions after hospitalization: a pilot prospective cohort study.

PURPOSE: Patients leaving the hospital are at increased risk of functional decline and hospital readmission. The Employee and Community Health service at Mayo Clinic in Rochester developed a care transition program (CTP) to provide home-based care services for medically complex patients. The study objective was to determine the relationship between CTP use, 30-day hospital readmission, and Emergency Room (ER) visits for adults over 60 years with high Elder Risk Assessment scores. PATIENTS AND METHODS: This was a pilot prospective cohort study that included 20 patients that used the CTP and 20 patients discharged from the hospital without using the CTP. The medically complex study patients were drawn from the department of Employee and Community Health population between October 14, 2011 and September 27, 2012. The primary outcomes were 30-day hospital readmission or ER visit after discharge from the hospital. The secondary outcomes were within-group changes in grip strength, gait speed, and quality of life (QOL). Patients underwent two study visits, one at baseline and one at 30 days postbaseline. The primary analysis included time-to-event from baseline to rehospitalization or ER visit. Paired t-tests were used for secondary outcomes, with continuous scores. RESULTS: Of the 40 patients enrolled, 36 completed all study visits. The 30-day hospital readmission rates for usual care patients were 10.5% compared with no readmissions for CTP patients. There were 31.6% ER visits in the UC group and 11.8% in the CTP group (P = 0.37). The secondary analysis showed some improvement in physical QOL scores (pre: 32.7; post: 39.4) for the CTP participants (P < 0.01) and no differences in gait speed or grip strength. CONCLUSION: Based on this pilot study of care transition, we found nonsignificant lower hospital and ER utilization rates and improved physical QOL scores for patients in the CTP group. However, the data leads us to recommend future studies with larger sample sizes (N = 250).

Predictors and costs of surgical site infections in patients with endometrial cancer.

OBJECTIVE: Technological advances in surgical management of endometrial cancer (EC) may allow for novel risk modification in surgical site infection (SSI). METHODS: Perioperative variables were abstracted from EC cases surgically staged between January 1, 1999, and December 31, 2008. Primary outcome was SSI, as defined by American College of Surgeons National Surgical Quality Improvement Program. Counseling and global models were built to assess perioperative predictors of superficial incisional SSI and organ/space SSI. Thirty-day cost of SSI was calculated. RESULTS: Among 1369 EC patients, 136 (9.9%) had SSI. In the counseling model, significant predictors of superficial incisional SSI were obesity, American Society of Anesthesiologists (ASA) score >2, preoperative anemia (hematocrit <36%), and laparotomy. In the global model, significant predictors of superficial incisional SSI were obesity, ASA score >2, smoking, laparotomy, and intraoperative transfusion. Counseling model predictors of organ/space SSI were older age, smoking, preoperative glucose >110 mg/dL, and prior methicillin-resistant Staphylococcus aureus (MRSA) infection. Global predictors of organ/space SSI were older age, smoking, vascular disease, prior MRSA infection, greater estimated blood loss, and lymphadenectomy or bowel resection. SSI resulted in a $5447 median increase in 30-day cost. CONCLUSIONS: Our findings are useful to individualize preoperative risk counseling. Hyperglycemia and smoking are modifiable, and minimally invasive surgical approaches should be the preferred surgical route because they decrease SSI events. Judicious use of lymphadenectomy may decrease SSI. Thirty-day postoperative costs are considerably increased when SSI occurs.

Primary care utilization and mental health diagnoses among adult patients requiring interpreters: a retrospective cohort study.

BACKGROUND: Patients requiring interpreters may utilize the health care system differently or more frequently than patients not requiring interpreters; those with mental health issues may be particularly difficult to diagnose. OBJECTIVE: To determine whether adult patients requiring interpreters exhibit different health care utilization patterns and rates of mental health diagnoses than their counterparts. DESIGN: Retrospective cohort study examining patient visits to primary care (PC), express care (EC), or the emergency department (ED) of a large group practice within 1 year. PATIENTS: Adult outpatients (n = 63,525) with at least one visit within the study interval and information regarding interpreter need. MAIN MEASURES: Mean visit counts, counts of mental disorders, and somatic symptom diagnoses between patients requiring interpreters (IS patients) and not requiring interpreters (non-IS patients). KEY RESULTS: IS patients (n = 1,566) had a higher mean number of visits overall (3.10 vs. 2.52), in PC (2.54 vs. 1.95), and in ED (0.53 vs. 0.44) than non-IS patients (all p < 0.01). IS patients had a lower mean number of visits in EC than non-IS patients (0.03 vs. 0.13; p < 0.01). Interpreter need remained a significant predictor of visit count in multivariate analyses including age, sex, insurance, and clinical complexity. A greater proportion of IS patients were high utilizers (10+ visits) than non-IS patients (3.6 % vs. 1.7 %; p < 0.01). IS patients had a lower frequency of mental health diagnoses (13.9 % vs. 16.7 %), but a higher frequency of diagnoses recognized as potential somatic symptoms including diseases of the nervous (29.3 % vs. 24.2 %), digestive (22.6 % vs. 14.5 %), and musculoskeletal systems (43.2 % vs. 34.5 %), and ill-defined conditions (61 % vs. 49.9 %), all p < 0.01. CONCLUSIONS: IS patients visited PC more often than their counterparts and were more often high utilizers of care. Two sources of high utilization, mental health diagnoses and somatic symptoms, differed appreciably between our populations and may be contributing factors.

Factors predictive of postoperative morbidity and cost in patients with endometrial cancer.

OBJECTIVE: To identify patient characteristics and perioperative factors predictive of 30-day morbidity and cost in patients with endometrial carcinoma. METHODS: Data of consecutive patients treated with hysterectomy for endometrial carcinoma between 1999 and 2008 were collected prospectively. Thirty predictors were chosen from more than 130 collected based on anticipated clinical relevance and prevalence (more than 3%). Complications were graded per the Accordion Classification. Multivariable models were developed using stepwise and backward variable selection methods. Thirty-day cost analyses were expressed in 2010 Medicare dollars. RESULTS: Of 1,369 patients, significant predictors (P<.01) of grade 2 and higher morbidity included American Society of Anesthesiologists physical status classification system class higher than 2 (odds ratio [OR] 2.1), preoperative white blood count (OR 2.1 per doubling), history of deep vein thrombosis (OR 2.1), pelvic and para-aortic lymphadenectomy (OR 2.3 compared with no lymphadenectomy), laparotomy (OR 2.8 compared with minimally invasive surgery), myometrial invasion more than 50% (OR 2.4), operating time (OR 1.9 per doubling), and grade 4 surgical complexity (OR 2.7 compared with grade 1). After controlling for patient factors in a multivariable model, laparotomy, pelvic, and para-aortic lymphadenectomy were associated with significant increases in cost compared with the use of minimally invasive surgery or hysterectomy alone. CONCLUSION: This analysis identifies patient and perioperative care factors predictive of 30-day morbidity and cost. These data are useful for preoperative counseling, for defining equitable reimbursement and factors critical for risk-adjustment when comparing outcomes, and for identifying areas for quality improvement in patients with endometrial carcinoma. Given the marked increases in morbidity and cost associated with laparotomy and lymphadenectomy, minimally invasive surgery and selective lymphadenectomy should be standard treatment for patients with endometrial carcinoma.

Getting physicians to open the survey: little evidence that an envelope teaser increases response rates.

BACKGROUND: Physician surveys are an important tool to assess attitudes, beliefs and self-reported behaviors of this policy relevant group. In order for a physician to respond to a mailed survey, they must first open the envelope. While there is some evidence that package elements can impact physician response rates, the impact of an envelope teaser is unknown. Here we assess this by testing the impact of adding a brightly colored "$25 incentive" sticker to the outside of an envelope on response rates and nonresponse bias in a survey of physicians. METHODS: In the second mailing of a survey assessing physicians' moral beliefs and views on controversial health care topics, initial nonrespondents were randomly assigned to receive a survey in an envelope with a colored "$25 incentive" sticker (teaser group) or an envelope without a sticker (control group). Response rates were compared between the teaser and control groups overall and by age, gender, region of the United States, specialty and years in practice. Nonresponse bias was assessed by comparing the demographic composition of the respondents to the nonrespondents in the experimental and control condition. RESULTS: No significant differences in response rates were observed between the experimental and control conditions overall (p = 0.38) or after stratifying by age, gender, region, or practice type. Within the teaser condition, there was some variation in response rate by years since graduation. There was no independent effect of the teaser on response when simultaneously controlling for demographic characteristics (OR = 0.875, p = 0.4112). CONCLUSIONS: Neither response rates nor nonresponse bias were impacted by the use of an envelope teaser in a survey of physicians in the United States.

Deployment of a mixed-mode data collection strategy does not reduce nonresponse bias in a general population health survey.

OBJECTIVE: To assess nonresponse bias in a mixed-mode general population health survey. DATA SOURCES: Secondary analysis of linked survey sample frame and administrative data, including demographic and health-related information. STUDY DESIGN: The survey was administered by mail with telephone follow-up to nonrespondents after two mailings. To determine whether an additional mail contact or mode switch reduced nonresponse bias, we compared all respondents (N = 3,437) to respondents from each mailing and telephone respondents to the sample frame (N = 6,716). PRINCIPAL FINDINGS: Switching modes did not minimize the under-representation of younger people, nonwhites, those with congestive heart failure, high users of office-based services, and low-utilizers of the emergency room but did reduce the over-representation of older adults. CONCLUSIONS: Multiple contact and mixed-mode surveys may increase response rates, but they do not necessarily reduce nonresponse bias.

Who doesn't authorize the linking of survey and administrative health data? A general population-based investigation.

PURPOSE: To determine the extent of authorization bias in a study linking survey and medical record data in a general population-based investigation. METHODS: Authorization status (authorized data linkage vs. not) was ascertained through a sequential mixed mode mail and telephone survey conducted in Olmsted County, MN. Respondents (regardless of authorization status) were linked to the Rochester Epidemiology Project (REP), the medical record system for health care providers in Olmsted County. The REP provided data on gender, age, race, health status (co-morbid conditions), and health care utilization (ER admission, hospital admission, clinical office visits and procedures). Authorizers (n=1357) are compared to non-authorizers (n=217) with respect to these demographic and clinical characteristics. RESULTS: 86.2% of respondents authorized data linkage. Non-authorizers were younger, healthier (lower Charlson score), and less likely to have 3 or more recent clinical office visits. In multivariate analysis, Charlson score was no longer a significant predictor of authorization while an ER visit did predict authorization. CONCLUSIONS: Younger subjects are less likely to authorize data linkages. As such, researchers should be aware of this source of potential bias when analyzing population-based linked survey and administrative data. The presence of bias with respect to health care use is more complicated. It is dependent on how the concept is operationalized with heavy clinical users more likely to authorize and those with ER visits less so.

Health Insurance Portability and Accountability Act (HIPAA) authorization and survey nonresponse bias.

OBJECTIVES: To extend earlier work (Beebe et al, Med Care. 2007;45:959-965) that demonstrated Health Insurance Portability and Accountability Act authorization form (HAF) introduced potential nonresponse bias (toward healthier respondents). RESEARCH DESIGN: The sample frame from the earlier experiment was linked to administrative medical record data, enabling the comparison of background and clinical characteristics of each set of respondents (HAF and No HAF) to the sample frame. SUBJECTS: A total of 6939 individuals residing in Olmsted County, Minnesota who were mailed a survey in September 2005 assessing recent gastrointestinal symptoms with an embedded HAF experiment comprised the study population. MEASURES: The outcomes of interest were response status (survey returned vs. not) by HAF condition (randomized to receive HAF or not). Sociodemographic indicators included gender, age, and race. Health status was measured using the severity-weighted Charlson Score and utilization was measured using emergency room visits, hospital admissions, clinic office visits, and procedures. RESULTS: Younger and nonwhite residents were under-represented and those with more clinical office visits were over-represented in both conditions. Those responding to the survey in the HAF condition were significantly more likely to be in poor health compared with the population (27.3% with 2+ comorbidities vs. 24.6%, P=0.02). CONCLUSIONS: The HAF did not influence the demographic composition of the respondents. However, in contrast to earlier findings based on self-reported health status (Beebe et al, Med Care. 2007;45:959-965), responders in the HAF condition were slightly sicker than in the non-HAF condition. The HAF may introduce a small amount of measurement error by suppressing reports of poor health. Furthermore, researchers should consider the effect of the HAF on resultant precision, respondent burden, and available financial resources.