Middle class populations have supported shared medical records, including Internet-accessible medical records. The attitudes of lower income populations, and of physicians, are less clear.
The objective of this study was to compare the attitudes toward shared outpatient medical records among (1) socioeconomically disadvantaged patients in community health centers, (2) insured patients in primary care offices, and (3) a broad range of physicians in outpatient practice.
Written questionnaires were provided to patients in the waiting rooms of six primary care practices in the metropolitan Denver, Colorado area. Three practices were community health centers, and three practices were primary care clinics of an academic medical center. Questionnaires were also mailed to primary care physicians in the state of Colorado.
There was a 79% response rate for patient surveys (601 surveys returned) and a 53% response rate for physician surveys (564 surveys returned). Academic medical center patients and community health center patients were equally likely to endorse shared medical records (94% vs 96%) and Internet-accessible records (54% vs 57%). Community health center patients were more likely than academic medical center patients to anticipate the benefits of shared medical records (mean number of expected benefits = 7.9 vs 7.1,
Interest in shared medical records is not confined to a white, middle class population. Shared medical records are almost universally endorsed across a broad range of ethnic and socioeconomic groups. A majority of patients are also interested in Internet-accessible records, but a substantial minority is not. The primary determinants of support of Internet-accessible records are not age, race, or education level; rather, they are previous experience with the Internet and patients' expectations of the benefits and drawbacks of reading their medical records. Physicians have more concerns about shared medical records and see less potential for benefit. The attitudes of patients and physicians may need to be reconciled for widespread adoption of shared medial records to be achieved.
As the Health Insurance Portability and Accountability Act (HIPAA) has clarified the rights of patients to review their medical records [
Several recent surveys have evaluated patients' interest in shared records. A study of patients in Minnesota found that 79% of patients were “very interested” or “somewhat interested” in reading their clinic medical record [
We addressed these issues through two related survey projects. In one project, we assessed the attitudes of a broad sample of physicians in the state of Colorado using a mailed questionnaire. In a follow-up project, we assessed patient attitudes in multiple primary care offices in the metropolitan Denver area. Half of these offices were associated with a community health center for socioeconomically disadvantaged patients, and half were primary care clinics of an academic medical center that provided services for a more middle class clientele. Our objectives were to compare the attitudes of patients in the two groups and to compare the attitudes of patients as a whole to those of doctors in the region.
Physician and patient questionnaires included demographic items and 16 questions assessing the potential benefits and concerns of sharing medical records (
Patients also answered two additional questions regarding their attitudes about shared medical records in general, and two additional questions regarding shared
The survey population represented outpatients to primary care practices in metropolitan Denver, Colorado. The sample frame consisted of adult patients (18 years of age and older) presenting for outpatient appointments to one of six primary care sites between September 1, 2003 and April 27, 2004. Three primary care practices associated with a teaching hospital (University of Colorado Hospital) represented patients typical of a private practice. Three neighborhood community health centers associated with the safety-net hospital (Denver Health) represented a financially disadvantaged and ethnically diverse population. A convenience sample was obtained from patients in the waiting rooms of these practices. All patients with appointments were potentially eligible. Because the medical records were written in English and we intended to study the attitudes of patients who would be reading their own medical records, patients who did not speak English were not approached for the survey.
Questionnaires were given to patients by a research assistant stationed in the waiting rooms of the practices. Surveys were anonymous, but the research assistant tracked how many patients declined to complete the survey. Surveys were abstracted and double-entry verified.
The survey of physicians was performed in July 2002. The survey population represented Colorado physicians in primary care (family practice, general internal medicine, and general practice) and in internal medicine specialties. The sample frame was derived from a list of Colorado physicians supplied by the Colorado Commission on Family Medicine. The original sample frame contained 4351 physician records with information on degree, specialty, age, gender, and street address. The database was cleaned to eliminate specialties not of interest to this study (615), duplicate entries (417), and entries without the full complement of information (50 due to missing age information, 6 due to missing gender information). This resulted in a cleaned database containing 3263 records. A probability sample was created by randomly selecting one fourth of the physicians in the primary care group and one half of the physicians in the internal medicine specialty group. This produced a sample of 1059 physicians, 580 in primary care and 479 in internal medicine specialties.
Questionnaires were mailed to physicians in July 2002. Potential respondents were initially mailed a postcard describing the survey. A written questionnaire was mailed one week later with a business reply envelope. A reminder card was sent two weeks later. A second questionnaire was mailed to those who did not respond within four weeks.
Statistical analyses were performed using SAS Version 9.1 (SAS Institute, Cary, NC). Differences in dichotomous outcomes were compared using chi-square tests, and differences in continuous outcomes were compared using
For patients, 601 surveys were returned, 295 from the community health centers (response rate 71%) and 306 from the academic primary care clinics (response rate 88%). For physicians, 340 questionnaires were returned from the primary care group (response rate 59%) and 224 from the specialist group (response rate 47%).
The majority of respondents in both patient groups were female, with a mean age in the 40s (
Patient demographics
|
|
|
|
Age (years), mean (SD) | 42 (15) | 49 (18) | < .001 |
Male gender | 75 (28) | 108 (37) | .02 |
White, non-Hispanic | 95 (35) | 222 (75) | < .001 |
Household income > $45000 per year | 16 (6) | 145 (52) | < .001 |
College graduate | 53 (20) | 165 (56) | < .001 |
Insurance other than Medicaid, Medically Indigent, or uninsured | 57 (22) | 263 (89) | < .001 |
More than three physician visits per year | 120 (41) | 95 (31) | .01 |
Used Internet before | 182 (67) | 241 (82) | < .001 |
Have Internet access at home or work | 148 (54) | 242 (83) | < .001 |
Interested in communicating with doctor by email | 129 (48) | 190 (66) | < .001 |
Reviewed parts of their medical records before | 131 (48) | 187 (63) | < .001 |
For physicians, the age and gender distribution of the respondent sample was representative of the sample frame (
Physician demographics
|
|
Age (years), mean (SD) | 48 (10) |
Male | 421 (75) |
Office-based practice | 535 (97) |
Already routinely send notes to patients | 45 (8) |
The responses of patients in the community health centers were compared with those from patients in the academic primary care clinics. Because the responses to the nine questions about potential benefits of access to the medical record were highly correlated (Cronbach alpha = 0.90), the count of the number of questions which were answered “strongly agree” or “agree” was created, which we termed the number of expected benefits.
In general, the patients in the community health centers (CHCs) were more likely to anticipate benefits (
Patients were also asked two summary questions about shared records, in general, and about shared records online, in particular. Ninety-five percent of all patients agreed with the statement, “Overall, I think it is a good idea for patients to be able to routinely review their outpatient medical records” (96% of CHC patients vs 94% of academic primary care clinic patients,
Expected benefits of shared medical records
Would improve understanding of medical conditions | 263 (90) | 249 (82) | .01 | 512 (86) | 220 (40) | < .001 |
Would improve understanding of doctors' instructions | 258 (89) | 230 (76) | < .001 | 488 (83) | 290 (53) | < .001 |
Would improve patient adherence | 255 (90) |
216 (72) | < .001 | 471 (81) | 257 (47) | < .001 |
Would prepare patients for visits |
253 (86) |
243 (80) | .04 | 496 (83) | 209 (38) | < .001 |
Would be reassuring |
258 (90) |
257 (85) | .06 | 515 (88) | 260 (47) | < .001 |
Would increase patients' sense of control |
263 (91) |
252 (83) | .003 | 515 (87) | 388 (70) | .001 |
Would increase trust in doctors |
242 (83) |
223 (75) | .02 | 465 (79) | 279 (52) | < .001 |
Would increase patient satisfaction |
254 (89) | 244 (82) |
.01 | 498 (85) | 240 (44) | < .001 |
Patients would identify errors in the medical record | 231 (83) | 253 (85) | .55 | 484 (84) | 177 (32) | < .001 |
Number of expected benefits, mean (SD) |
7.9 (2.0) |
7.1 (2.6) |
< .001 | 7.5 (2.3) | 4.2 (3.0) | < .001 |
Other expectations of shared medical records
Lab and x-ray reports would be confusing |
146 (50) |
109 (36) | < .001 | 255 (43) | 421 (76) | < .001 |
Doctors' notes would be confusing |
130 (44) |
84 (28) | < .001 | 214 (36) | 274 (49) | < .001 |
Would increase patient worry |
84 (29) |
68 (22) | .07 | 152 (26) | 448 (81) | < .001 |
Would cause offense or embarrassment |
55 (19) |
29 (10) | < .001 | 84 (14) | 248 (45) | < .001 |
Would increase questions between visits |
198 (69) |
142 (47) | < .001 | 340 (58) | 385 (70) | < .001 |
To assess the determinants of patient attitudes towards Internet-accessible medical records, we created a logistic model. The dependent (outcome) variable was agreement with the statement, “Overall, I think it is a good idea for patients to be able to review their outpatient medical records using the Internet.” Bivariate analyses were performed and demographic variables (listed in
Previous use of the Internet (OR = 2.45, CI 1.59–3.79)
The number of expected benefits (OR = 1.12 per question, CI 1.03–1.21). The mean number of expected benefits for those who endorsed Internet-accessible records was 7.8 vs 7.1 for those who did not endorse them.
Anticipating asking more questions between visits (OR = 1.73, CI 1.18–2.54)
Anticipating doctors' notes being confusing (OR = 1.50, CI 1.01–2.22)
The patient responses in aggregate were compared with the responses from the physician survey. Of note, the responses of primary care and specialist physicians were combined, as were responses of patients at the community health centers and the academic clinics, since the differences between patients and physicians was much greater than the differences within physician and patient subgroups. Because the inter-item correlations of the expected benefits was also high among physicians (Cronbach alpha = 0.87), we used the number of expected benefits as for patients.
Physicians were significantly more likely to anticipate concerns than patients (
Physicians were also asked two additional questions about their expectations if patients could routinely review their outpatient medical records. Sixty-three percent anticipated that their “workload would increase substantially,” and 45% anticipated that they “would document things differently in the medical record.”
This survey confirms the primary results of the surveys in Minnesota [
Our survey also extends these findings by comparing patient attitudes to the significantly different attitudes of physicians. Patients are particularly likely to anticipate that shared records will be empowering, and particularly unlikely to anticipate that access to their medical records will be embarrassing. Physicians, by contrast, are especially likely to anticipate that laboratory results will confuse patients and that shared records will make patients worry more.
In addition to our quantitative findings, our anecdotal experience in conducting the survey confirmed the strong polarity of opinion towards Internet-accessible records that was reported in the Minnesota survey [
Our results also complement the findings of Hassol et al in the Geisinger Health System in Pennsylvania [
In addition, Hassol's study reported that Geisinger physicians and system administrators expressed particular concern that patients would be worried about test results that were available online. This information was only anecdotal, however, because their clinician response rate (13%) was too low for statistical analysis. The larger response rate in our statewide physician survey confirms that the majority of physicians are concerned about the potential for shared medical records to confuse or worry their patients.
The different sampling strategies we used for the physician and the patient surveys appear to have been successful in obtaining a representative response of the populations. The response rate for the physician survey is typical of mailed surveys of physicians [
Several limitations of this study are noted. The attitudes of Colorado physicians and metropolitan Denver patients are only incomplete representations of broader national opinions. Because the patient and the physician surveys were conducted over a year apart, secular changes in attitudes may have affected the comparisons. Also, while the questions in the patient survey and the physician survey were linked, the differences in the way the questions were framed may have accounted for some of the differences observed in the physician and the patient responses.
Overall, our survey confirms that nearly all patients value having access to their medical records. Clearly, patient-accessible medical records are not something valued only by a privileged elite or by patients with idiosyncratic relationships with the medical system [
This study was funded in part by a grant from the Commonwealth Fund (grant number 20010374) and by a grant from the University of Colorado Division of General Internal Medicine Small Grants Program. These agencies did not require review or approval of the manuscript.
John Steiner MD provided expert consultation in the design and coordination of the project. Andy Steele MD, MPH provided oversight of the survey process in the Denver health system. The survey of patients was conducted under the auspices of CaReNet, a practice-based research network of primary care practices in the state of Colorado. The database of physicians was made available by the Colorado Commission of Family Medicine.
The authors have no financial interests in patient-accessible medical records. Dr. Ross and Dr. Lin practice at University of Colorado Hospital, which has sponsored pilot programs of online patient-accessible medical records.
Overview of questions asked in physician and patient questionnaires assessing the potential benefits and concerns of sharing medical records.
Actual physician questionnaire.
Actual patient questionnaire.
community health center
odds ratio