Potential Effect of Authorization Bias on Medical Record Research

doi:10.4065/74.4.330

Mayo Clinic Proceedings

Volume 74, Issue 4, April 1999, Pages 330-338

https://doi.org/10.4065/74.4.330 Get rights and content

Objective

To analyze the influence of recent changes in Minnesota statutes that generally require prior authorization for use of medical records for research from patients who received medical care after Jan. 1,1997.

Material and Methods

In this Mayo Clinic Institutional Review Board-approved study, we obtained a stratified random sample of patients encountered at Mayo Clinic Rochester during the period 1994 through 1996 and estimated the proportion willing to provide the general authorization. On the basis of data from administrative files, we then compared demographic, diagnostic, and utilization characteristics for patients who provided authorization and those who did not.

Results

Overall, 3.2% (95% confidence interval, 2.4 to 4.0%) of the study subjects declined authorization. If patients not responding to requests for authorization were also considered to have refused, the overall refusal rate would be 20.7% (95% confidence interval, 18.5 to 22.9%). Women were somewhat more likely to refuse authorization than were men (4.0% versus 2.4%; P = 0.067), and patients younger than 60 years were more likely to refuse than were older patients (5.4% versus 1.2%; P<0.001). Patients residing more than 120 miles from Rochester were much less likely to decline authorization than were local residents (2.1% versus 5.8%; P = 0.001). Patients with prior diagnoses that might be considered more sensitive such as mental disorders, infectious diseases, and reproductive problems also were more likely to refuse authorization.

Conclusion

These data demonstrate that laws requiring written authorization for research use of the medical record could result in substantial biases in etiologic and outcome studies, the direction and magnitude of which may vary from topic to topic. Clinicians should be prepared to enter the discussion to help inform patients and legislators of the potential hazards of laws that restrict access to medical records for research purposes.

Section snippets

Study Setting

Mayo Clinic is best known as a major tertiary referral center, but Mayo Clinic Rochester also provides comprehensive care for the region in every clinical discipline, including a major portion of the primary care received by the residents of Olmsted County, Minnesota. The two Mayo Clinic-affiliated hospitals in Rochester—Saint Marys Hospital and Methodist Hospital—have a combined total of 1,900 beds. In 1995, there were almost 250,000 patient registrations at Mayo Clinic Rochester-67,100

RESULTS

Altogether, 2,023 of 2,463 potential subjects (82.6% with sampling weights) responded to our request for the research authorization. Of these, 1,408 subjects (57.9%) responded to the initial request, whereas 330 responded to the second letter (30.6% of the 1,055 subjects remaining) and 285 responded to the third letter (40.3% of the 725 subjects remaining). The refusal (nonauthorization) rate increased from 2.7% of those returning the first letter to 5.2% of patients responding to the second

DISCUSSION

In this study, only 3.2%.—of patients explicitly exercised their right to privacy and refused to.—provide a broad authorization that would permit use of medical record data for research purposes. This authorization covers any study approved by the Mayo Clinic Institutional Review Board in perpetuity, although patients have the right to withdraw the authorization at a later date if they wish. This finding corresponds to the general Mayo Clinic experience as of October 1998, in which only 3.8% of

CONCLUSION

The data from this study suggest that the Minnesota law could have had an influence on the characteristics of medical records made available for research purposes. The effect on the conclusions of any study, however, probably cannot be deduced. Legislation requiring active written consent in each instance before review of medical records could be devastating and could result in an effective authorization refusal rate of 20 to 30%. The only assurance of valid results is a complete and unbiased

ACKNOWLEDGMENT

We thank Sondra L. Buehler for assistance in the preparation of the submitted manuscript. John La Forgia and Karel M. Weigel helped develop the cover letter, educational brochure, and authorization form based on input from the focus groups, which were critical in the development of this study protocol. The insightful comments of the masked reviewers were greatly appreciated.

REFERENCES (34)

LJ Melton
History of the Rochester Epidemiology Project
Mayo Clin Proc
(1996)
ME Charlson et al.
A new method of classifying prognostic comor bidity in longitudinal studies; development and validation
J Chronic Dis
(1987)
RA Deyo et al.
Adapting a clinical comorbidity index for use with ICD-9CM administrative databases
J Clin Epidemiol
(1992)
LJ Melton et al.
Non-response bias in studies of diabetic complications: the Rochester Diabetic Neuropathy Study
J Clin Epidemiol
(1993)
LA Panser et al.
Effect of several recruitment strategies on response rates at baseline in a prospective cohort investigation: the Olmsted County Study of Urinary Symptoms and Health Status among Men
Ann Epidemiol
(1994)
KD Lindsted et al.
Healthy volunteer effect in a cohort study: temporal resolution in the Adventist Health Study
J Clin Epidemiol
(1996)
R Hébert et al.
Refusal and information bias associated with postal questionnaires and face-to-face interviews in very elderly subjects
J Clin Epidemiol
(1996)
BA Weiner et al.
Confidentiality of patient-related information
Arch Ophthalmol
(1994)
L Gordis et al.
Privacy, confidentiality, and the use of medical records in research
Science
(1980)
EA Codman
The product of a hospital
Surg Gynecol Obstet
(1914)

EA Codman et al.

Report of the Committee on Standardization of Hospitals

Surg Gynecol Obstet

(1914)

HM Connolly et al.

Valvular heart disease associated with fenfluramine-phentermine. N

Engl J Med

(1997)

SS Coughlin et al.

Ethics, scientific validity, and the design of epidemiologic studies

Epidemiology

(1992)

EG Krtox

Confidential medical records and epidemiological research

BMJ

(1992)

RL Berkelman et al.

Public health surveillance of non-infectious chronic diseases: the potential to detect rapid changes in disease burden

Int J Epidemiol

(1990)

Directive 95/46/EC of the European Parliament and of the Council of 24 October 1995

Official J Eur Communities

(1995)

U Cordier

The directive on the protection of individuals with regard to the processing of personal data, and medical epidemiological research

Eur Union Biomed Health Res

(1995)

Cited by (113)

Exercise Test Predicts Both Noncardiovascular and Cardiovascular Death in a Primary Prevention Population
2023, Mayo Clinic Proceedings
To identify specific causes of death and determine the prevalence of noncardiovascular (non-CV) deaths in an exercise test referral population while testing whether exercise test parameters predict non-CV as well as CV deaths.
Non-imaging exercise tests on patients 30 to 79 years of age from September 1993 to December 2010 were reviewed. Patients with baseline CV diseases and non-Minnesota residents were excluded. Mortality through January 2016 was obtained through Mayo Clinic Records and the Minnesota Death Index. Exercise test abnormalities included low functional aerobic capacity (ie, less than 80%), heart rate recovery (ie, less than 13 beats/min), low chronotropic index (ie, less than 0.8), and abnormal exercise electrocardiogram (ECG) of greater than or equal to 1.0 mm ST depression or elevation. We also combined these four abnormalities into a composite exercise test score (EX_SCORE). Statistical analyses consisted of Cox regression adjusted for age, sex, diabetes, hypertension, obesity, current and past smoking, and heart rate–lowering drug.
The study identified 13,382 patients (females: n=4736, 35.4%, 50.5±10.5 years of age). During 12.7±5.0 years of follow-up, there were 849 deaths (6.3%); of these 162 (19.1%) were from CV; 687 (80.9%) were non-CV. Hazard ratios for non-CV death were significant for low functional aerobic capacity (HR, 1.42; 95% CI, 1.19 to 1.69; P<.0001), abnormal heart rate recovery (HR, 1.36; 95% CI, 1.15 to 1.61; P<.0033), and low chronotropic index (HR, 1.49; 95% CI, 1.26 to 1.77; P<.0001), whereas abnormal exercise ECG was not significant. All exercise test abnormalities including EX_SCORE were more strongly associated with CV death versus non-CV death except abnormal exercise ECG.
Non-CV deaths predominated in this primary prevention cohort. Exercise test abnormalities not only predicted CV death but also non-CV death.
Serum 25-hydroxyvitamin D values and risk of incident cardiovascular disease: A population-based retrospective cohort study
2021, Journal of Steroid Biochemistry and Molecular Biology
Concentrations of serum 25-hydroxyvitamin D [25(OH)D] below 20 ng/mL and above 50 ng/mL have been associated with chronic adverse events including cardiovascular disease.
To conduct a comprehensive population-based study in the United States of the relationship of low and high serum 25(OH)D levels with cardiovascular disease.
We identified all serum 25(OH)D measurements in adults age 18 years and older residing in Olmsted County, MN between January 1, 2005 and December 31, 2011, using the resources of the Rochester Epidemiology Project. Any new diagnosis of cardiovascular disease was the primary outcome, and time zero was 30 days after first 25(OH)D measurement. Patients were followed until their last clinical visit as an Olmsted County resident, December 31, 2014, or death. Categories of 25(OH)D values were examined using predetermined ranges of interest: <12, 12−19, 20−50 (reference range), and >50 ng/mL. Multivariable Cox proportional hazards models were adjusted for age, BMI, sex, race, smoking history, season of 25(OH)D measurement, hypertension, hyperlipidemia, socioeconomic status and Charlson comorbidity index at time of 25(OH)D measurement.
A total of 11,002 unique persons had a 25(OH)D measurement, with a mean (±SD) value of 30.0 ± 12.9 ng/mL. Mean age was 54.3 ± 17.2 years, and the majority were female (77.1 %) and white (87.6 %). There were 4124 new diagnoses of cardiovascular disease in this cohort after a median overall follow-up of 4.8 years (IQR 3.4–6.2). Adjusted cardiovascular disease hazard ratios (HRs) (95 % confidence interval) for 25(OH)D values <12, 12−19, and >50 ng/mL, compared to the reference range 20−50 ng/mL, were 1.28 (1.12–1.46), 1.19 (1.09–1.31), and 1.10 (0.95–1.26), respectively.
Values of 25(OH)D <20 ng/mL were associated with development of a new diagnosis of cardiovascular disease. There was no significant association between 25(OH)D values >50 ng/mL and cardiovascular disease.
The relationship of 25-hydroxyvitamin D concentrations and individual-level socioeconomic status
2020, Journal of Steroid Biochemistry and Molecular Biology
Socioeconomic status (SES), defined as the ability to access desired resources, is associated with behaviors that may affect vitamin D status. Most studies of the effect of vitamin D status on outcomes do not account for individual-level SES. The ability to adjust for SES in epidemiologic studies, when data on conventional SES measures have not been obtained, would be advantageous. We identified all serum 25(OH)D measurements in adults age 18 years and older residing in Olmsted County, MN, a mixed urban-rural setting, between January 1, 2005 and December 31, 2011, through the Rochester Epidemiology Project. The first 25(OH)D measurement was considered the index measurement for each subject. SES was determined for each subject by the HOUsing-based SocioEconomic Status (HOUSES) index, derived from real property data. The HOUSES index is an aggregated z-score of assessed housing value, area of living space, number of bedrooms, and number of bathrooms, with higher scores indicating higher SES. Multivariable analyses were adjusted for age, BMI, sex, race, season of 25(OH)D measurement, and Charlson comorbidity index. HOUSES was matched for 10,378 of 11,002 subjects (94 %) with 25(OH)D measurements available. The mean (SD) age was 54.3 (17.1) years with 26.9 % ≥65 years; 77.3 % were women, and 12.1 % were non-white. The mean 25(OH)D concentration was 30.0 (12.9) ng/mL, and 598 (5.8 %) had a 25(OH)D value <12 ng/mL. The mean (SD) HOUSES was -1.55 (3.09),-0.97 (3.34), 0.14 (3.52), 0.24 (3.51) for serum 25(OH)D categories of <12, 12–19, 20–50, and >50 ng/mL, respectively (P = 0.12 for trend). 25(OH)D increased by 0.43 (95 % CI 0.36-0.50) ng/mL for each unit increase in HOUSES in univariate analysis and by 0.28 (0.21-0.35; P < 0.001) ng/mL in multivariable analysis. This represents a change of 4 ng/mL across the entire range of observed HOUSES, an effect similar in magnitude to the seasonal variation of 25(OH)D values. SES was independently associated with serum 25(OH)D concentrations in a dose-response manner after adjustment for important covariates. HOUSES is a useful tool to assess the role of individual-level SES in health outcomes when other SES measures are unavailable and to control for confounding by SES in examining the effect of 25(OH)D on clinical and metabolic outcomes.
Cumulative Doses of Ionizing Radiation From Computed Tomography: A Population-Based Study
2019, Mayo Clinic Proceedings
To assess cumulative radiation doses from computed tomography (CT), patient characteristics, and clinical indications for CT in a population-based sample.
A cohort study using medical records linkage through the Rochester Epidemiology Project was conducted to ascertain all CT examinations in Olmsted County, Minnesota, performed between January 1, 2004, and December 31, 2013, among all adults who were alive for 3 or more years after the end of follow-up (to exclude exposures preceding death). Ten-year cumulative effective ionizing radiation doses were estimated on the basis of typical doses per CT modality. Among patients with high doses (≥100 mSv/10 years), CT scans were reviewed for clinical setting, indications, and results.
Of 54,447 adults (median age, 44.0 years at inclusion), 26,377 (48.4%) underwent at least one CT. Ten-year radiation doses from CT were 0.1 to 9.9 mSv in 15.8% of the population (8593 patients), 10 to 24.9 mSv in 16.9% (9502), 25 to 99.9 mSv in 13.8% (7492), and 100 mSv or greater in 1.9% (1041). Computed tomography of the abdomen and pelvis accounted for 67.2% of the estimated dose. In multivariable models, doses differed 1.21-fold to 2.16-fold between extreme categories of age, body mass index, education level, smoking status, and by race. Of 600 CTs in 200 patients with high doses, 70.5% were obtained for restaging of solid cancers and lymphoma, abdominal pain, infection, kidney stones, follow-up of nodules or masses, and chest pain/evaluation for pulmonary embolism.
Exposure to ionizing radiation from CT occurred disproportionally in specific subgroups of the population. A limited number of clinical indications contributed the majority of radiation among adults with high doses.
Population-Based Incidence of Potentially Life-Threatening Complications of Hypocalcemia and the Role of Vitamin D Deficiency
2019, Journal of Pediatrics
To determine the incidence of potentially life-threatening complications of hypocalcemia in infants and children in Olmsted County, Minnesota; and to determine if vitamin D deficiency contributed to these events and was, at the time of clinical presentation, considered as a possible cause.
In this population-based descriptive study, data were abstracted from the Rochester Epidemiology Project, a medical record linkage system covering 95% of patients in Olmsted County, Minnesota. Participants were children aged 0-5 years who resided in Olmsted County between January 1, 1996 and June 30, 2017, and who received diagnoses of seizures, cardiomyopathy, cardiac arrest, respiratory arrest, laryngospasm, and/or tetany. The incidence of hypocalcemia plus a potentially life-threatening complication was calculated.
Among 15 419 patients aged 0-5 years in Olmsted County during the study period, 1305 had eligible complications: 460 had serum calcium checked within 14 days of presentation and 85 had hypocalcemia. Patients were excluded when causes other than hypocalcemia likely triggered the complication, leaving 16 children whose complication was attributed to hypocalcemia. Three of these 16 patients had a serum 25-hydroxyvitamin D measurement and 2 were deficient (≤6 ng/mL [15 nmol/L]). Among children aged 0-5 years, the incidence of hypocalcemia plus a potentially life-threatening complication was 6.1 per 100 000 person-years (95% CI, 3.5-10.0).
Vitamin D deficiency is an underinvestigated cause of complications of hypocalcemia in children. Serum calcium and 25-hydroxyvitamin D should be measured in children with these complications to identify possibly life-threatening vitamin D deficiency.
Serum 25-Hydroxyvitamin D Values and Risk of All-Cause and Cause-Specific Mortality: A Population-Based Cohort Study
2018, Mayo Clinic Proceedings
To determine the relationship between 25-hydroxyvitamin D (25[OH]D) values and all-cause and cause-specific mortality.
We identified all serum 25(OH)D measurements in adults residing in Olmsted County, Minnesota, between January 1, 2005, and December 31, 2011, through the Rochester Epidemiology Project. All-cause mortality was the primary outcome. Patients were followed up until their last clinical visit as an Olmsted County resident, December 31, 2014, or death. Multivariate analyses were adjusted for age, sex, race/ethnicity, month of measurement, and Charlson comorbidity index score.
A total of 11,022 individuals had a 25(OH)D measurement between January 1, 2005, and December 31, 2011, with a mean ± SD value of 30.0±12.9 ng/mL. Mean age was 54.3±17.2 years, and most were female (77.1%) and white (87.6%). There were 723 deaths after a median follow-up of 4.8 years (interquartile range, 3.4-6.2 years). Unadjusted all-cause mortality hazard ratios (HRs) and 95% CIs for 25(OH)D values of less than 12, 12 to 19, and more than 50 ng/mL were 2.6 (95% CI, 2.0-3.2), 1.3 (95% CI, 1.0-1.6), and 1.0 (95% CI, 0.72-1.5), respectively, compared with the reference value of 20 to 50 ng/mL. In a multivariate model, the interaction between the effect of 25(OH)D and race/ethnicity on mortality was significant (P<.001). In white patients, adjusted HRs for 25(OH)D values of less than 12, 12 to 19, 20 to 50, and greater than 50 ng/mL were 2.5 (95% CI, 2.2-2.9), 1.4 (95% CI, 1.2-1.6), 1.0 (referent), and 1.0 (95% CI, 0.81-1.3), respectively. In patients of other race/ethnicity, adjusted HRs were 1.9 (95% CI, 1.5-2.3), 1.7 (95% CI, 1.1-2.6), 1.5 (95% CI, 1.0-2.0), and 2.1 (95% CI, 0.77-5.5).
White patients with 25(OH)D values of less than 20 ng/mL had greater all-cause mortality than those with values of 20 to 50 ng/mL, and white patients had greater mortality associated with low 25(OH)D values than patients of other race/ethnicity. Values of 25(OH)D greater than 50 ng/mL were not associated with all-cause mortality.

View all citing articles on Scopus

: This study was supported in part by Grant AR 30582 from the National Institutes of Health, Public Health Service.

View full text

Potential Effect of Authorization Bias on Medical Record Research

Objective

Material and Methods

Results

Conclusion

Section snippets

Study Setting

RESULTS

DISCUSSION

CONCLUSION

ACKNOWLEDGMENT

Mayo Clin Proc

J Chronic Dis

J Clin Epidemiol

J Clin Epidemiol

Ann Epidemiol

J Clin Epidemiol

J Clin Epidemiol

Confidentiality of patient-related information

Arch Ophthalmol

Privacy, confidentiality, and the use of medical records in research

Science

The product of a hospital

Surg Gynecol Obstet

Report of the Committee on Standardization of Hospitals

Surg Gynecol Obstet

Valvular heart disease associated with fenfluramine-phentermine. N

Engl J Med

Ethics, scientific validity, and the design of epidemiologic studies

Epidemiology

Confidential medical records and epidemiological research

BMJ

Public health surveillance of non-infectious chronic diseases: the potential to detect rapid changes in disease burden

Int J Epidemiol

Directive 95/46/EC of the European Parliament and of the Council of 24 October 1995

Official J Eur Communities

The directive on the protection of individuals with regard to the processing of personal data, and medical epidemiological research

Eur Union Biomed Health Res