Health information seeking. Error bars represent 95% confidence intervals. The asterisk indicates the reference group was "Internet not very or not at all important"; the dagger, the reference group was "last time used Internet for health information was more than 6 months, not sure, or never."
Relationship with physician. Error bars represent 95% confidence intervals. The asterisk indicates the reference group was "have faith/trust in primary physician—agree or strongly agree."
Health status. Error bars represent 95% confidence intervals. The asterisk indicates the reference group was "excellent general health."
Health care use. Error bars represent 95% confidence intervals. The asterisk indicates the reference group was "no visits"; the dagger, the reference group was "none or 1 provider."
Health habits. Error bars represent 95% confidence intervals. The asterisk indicates the reference group was "health habits somewhat better or about the same as other of same age."
Concern about health. Error bars represent 95% confidence intervals. The asterisk indicates the reference group was "not very or not at all worried"; the dagger, the reference group was "think about health rarely or never."
Patient safety. Error bars represent 95% confidence intervals. The asterisk indicates the reference group was "not very or not at all concerned about error"; the dagger, the reference group was "suffered no harm."
Record access. Error bars represent 95% confidence intervals.
Sociodemographic characteristics. Error bars represent 95% confidence intervals. The asterisk indicates the reference group was "date of birth before 1946."
Fowles JB, Kind AC, Craft C, Kind EA, Mandel JL, Adlis S. Patients' Interest in Reading Their Medical RecordRelation With Clinical and Sociodemographic Characteristics and Patients' Approach to Health Care. Arch Intern Med. 2004;164(7):793-800. doi:10.1001/archinte.164.7.793
Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004
Although opportunities for patients to review their medical records are increasing, nothing is known about which patients want to take advantage of those opportunities. The objective of this study was to determine the proportion and characteristics of patients who are very interested in examining their clinic medical record and the reasons for their interest.
Cross-sectional, mailed survey (conducted in May 2001) to a random sample of 4500 adults who had a recent clinic visit.
The response rate was 81%; 36% were very interested in reading their medical record (dependent variable). In multivariate logistic regression, the significantly related factors were seeking health information (finding the Internet very important for health information [adjusted odds ratio, 2.09], having a health newsletter subscription [adjusted odds ratio, 1.23], and using a health resource book in last month [adjusted odds ratio, 1.36]); being very concerned about errors in care (adjusted odds ratio, 2.52); and lacking trust in their physician (adjusted odds ratio, 1.55). Health status, use of health care, education, and income were not independently related to patients' interest. The most common reasons for patients wanting to look at their medical record were to see what their physician said about them (74%), to be more involved in their health care (74%), and to understand their condition better (72%).
Patients' interest in reading their medical record is better predicted by their consumer approach to health care than it is by their clinical characteristics. Demographic characteristics of sex and race were related, while socioeconomic factors of education and income were not.
The momentum to involve patients in all aspects of their health and health care is increasing. This movement is demonstrated in the drive to engage patients in managing their chronic diseases,1- 3 in selecting care based on quality,4 in improving the quality of the care delivery system,5,6 and in increasing their share of costs.7 Simultaneously, information sources that support patients' and consumers' active participation are rapidly expanding. A plethora of new materials and Web-based health sites support chronic disease self-management8- 10; compare the quality of physicians, hospitals, and health plans4- 11; and help consumers select among health coverage options.12
In this consumer-centered context, one information source has often been ignored—the patient's own medical record. The Institute of Medicine recognized this gap in its report Crossing the Quality Chasm,5 and recommended that "patients should have unfettered access to their own medical information." The federally enacted Health Insurance Portability and Accountability Act13 expanded and unified disparate state regulations to ensure that patients will have access to their records. The arrival of electronic medical records may mean that patients will have easier access to their medical records.14,15
Although patients may soon have ready access to their medical records, the nature of their interest and degree of their demand is unknown. To help plan for this era, we designed the current study to understand the characteristics of patients who would be interested in looking at their medical records.
In 1973, a seminal editorial by Shenkin and Warner16 began the debate of whether patients should be given their medical records. Over the next quarter century, subsequent editorials have debated the merits of releasing medical records to patients, some believing that trust would be eroded,17 that patients would become needlessly worried,18 or that physicians would be constrained in what they would record.19,20 On the other hand, numerous benefits have been suggested, including empowering patients,21 correcting inaccuracies and encouraging better record keeping,22- 24 and improving patient-physician communication and increasing trust.22,25- 27
Empirical research regarding the effect of sharing medical records with patients is limited and dated. The conclusions of these studies are often constrained by small sample sizes and weak study designs. Taken together, they suggest that sharing the medical record has a modest, positive effect on knowledge or perceived knowledge,28- 34 on patient-physician relationships,34- 37 and even on limited health outcomes.28,38- 41 We found only 1 survey that asked patients if they were interested in seeing their medical record. In a 1988 survey of 200 indigent patients in Alabama, 80% reported that patients should be allowed to see their medical record and 52% believed they should be given a copy to take home.42
We approached this study with 3 broad theories about the factors related to patients' interest in their medical record. First, we hypothesized that patients who were sick and used more health services would be more interested in reading their medical record than patients who were healthy. Second, we were interested in people's characteristics as health care consumers. We hypothesized that people with a greater interest in health (independent of the medical condition) and concern about health care quality would be more interested in reading their medical record than those with less interest. Finally, we predicted that interest would be a function of higher socioeconomic status and being a woman.
This survey was conducted at Park Nicollet Health Services, a not-for-profit care system with 550 physicians who deliver both primary and specialty care to approximately 17% of the population in Minneapolis, Minn, and its suburbs.
Using a random number table, we randomly selected 4500 adult patients (age ≥18 years) from the 39 008 patients who had billable encounters in April 2001. Sample size was based on the desired precision of the estimate as well as the ability to detect a 5% difference on various characteristics of those very interested and those less than very interested. We hypothesized that 15% of respondents would be very interested. The 95% confidence interval (precision of the estimate) for the questionnaire with 3150 responses (15% being very interested) was 13.8% to 16.2%. Assuming an α error of .05 (2-sided test), we had 97% power to detect the difference of 5%.
The 6-page, 57-item questionnaire included items from national surveys43- 48 and questions designed for this survey. The questionnaire included the following patient constructs: health status, health care use, health concern, health habits, relationship with personal physician, degree of involvement in health decision making, general information–seeking patterns, health information–seeking patterns, concern about medical errors, experience of medical injury, history of work in health care, history of reading medical record, and sociodemographic characteristics (age, sex, race and ethnicity, income, education, and employment status). The questionnaire also had items about which sections of the medical record patients would be interested in reviewing and a list of possible reasons why they would be interested. It incorporated a set of items addressing preferred modes of obtaining access to the medical record (paper or electronic; in the clinic or at home). We conducted cognitive testing on the new and previously used items on 8 patients. Two members of the research team asked patients in the clinic's laboratory waiting room to complete and critique the questionnaire. Minor formatting changes were made, and several items were reworded to make their meaning clearer. The investigators and their consultants again reviewed the items and the constructs of interest to assess face and content validity. The study protocol and questionnaire were reviewed for scientific merit by Park Nicollet Institute's Protocol Review Committee and for the protection of human subjects by its institutional review board.
A modified Dillman survey technique was used.49 Beginning in May 2001, we sent a series of 5 mailings to the patient sample: (1) a prenotification letter from the clinic medical director in week 1, (2) a cover letter, questionnaire, and $2 cash incentive in week 2, (3) a reminder postcard in week 3, (4) a second copy of the questionnaire to nonrespondents in week 5, and (5) a third copy of the questionnaire to remaining nonrespondents in week 7.
The main outcome was patients' self-reported interest in examining their own medical record: "How interested are you in reading your own medical record? (1) very interested, (2) somewhat interested, (3) uncertain, (4) not very interested, (5) not at all interested."
The dependent measure was collapsed into 2 categories: very interested in reading your own record and less than very interested. We focused on the uppermost value "very interested" because the patient satisfaction literature indicates that people are more likely to act on the uppermost value rather than a combination of the top 2 values.50 We compared the patients' interest in reading the medical record with each of the independent measures using contingency tables with the χ2 statistic. Development of a logistic regression model began with grouping variables representative of the following major constructs: health status, health care use, concerns about health and health care, relationship with physician, health information seeking, and sociodemographic characteristics. Variables representing a construct were used as the independent variables in separate logistic regression models. To be included in a construct group the variables had to be significant at the .05 level. The variables that were significant in each construct model were then entered into the full logistic model. Three disease conditions (asthma, migraine, and hay fever) that were not significant in the construct models were included in the full logistic model because of their clinical interest to the investigators. Adjusted odds ratios and 95% confidence intervals were calculated for each variable included in the final model. The goodness-of-fit of the final model was assessed using the Hosmer and Lemeshow test. The result indicated that the null hypothesis of good fit could not be rejected.
Forty-three questionnaires were undeliverable and were subtracted from the denominator. The response rate was 80.9% (3609/4457). Respondents were somewhat more likely to be women and to be older than nonrespondents. Sixteen of the returned questionnaires lacked responses to the dependent measure and could not be included in the analysis (n = 3593). Most of the respondents were women, well educated, and white (Table 1).
Over one third of the respondents (36%) were very interested in reading their clinic medical record. Another 43% were somewhat interested. At the bivariate level, many of the independent variables were significantly related to the patients' interest in reading the medical record (Table 1).
The results of the adjusted odds ratios for the full logistic model are shown in Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, Figure 8, and Figure 9. In this multivariate analysis, health status, and use of health care were not independently related to the patients' interest in reading the medical record (Figure 3 and Figure 4).
Various measures of health information seeking were independently related to being very interested in reading their medical record, including finding the Internet an important source of health information, having a health newsletter subscription, and using a health resource book in the last month (Figure 1). Lack of trust in their physician and being very concerned about patient errors remained related factors (Figure 2 and Figure 7).
Those who had looked at their medical record in the past were almost 3 times as likely to be very interested in reading their record than those who had not. Patients who did not know that they had the legal right to inspect their records were more than twice as likely to be very interested in reading their record (Figure 8).
Educational level and income were not independently related to the patients' interest in reading the medical record. Women were almost twice as likely as men to be very interested, and being nonwhite remained a significant independent predictor (Figure 9).
Respondents who were very interested in reading their medical record had many reasons, and most had more than 1 (Table 2). The most common reason was to be more active in their own health care. They wanted to see what their physician said about them and wanted to understand their condition better. Patients were least interested in trying to figure out what might be wrong with them; still, almost two thirds of respondents selected this reason.
We asked patients which specific portion of the medical record they were interested in seeing. Patients were most often interested in seeing their laboratory results (55.2%), closely followed by the physician's notes (53.3%). They were least interested in their past medications, but medications were still important to almost one third of respondents.
When asked how often they would like to look at their record, respondents reported that they wanted to read their medical records at regular, but not frequent, intervals. Almost half of the sample (48.3%) was interested in looking at their medical record once a year, 10.8% were interested in looking at it more often and 40.1% less often.
Only 11% strongly agreed that their physician would think it was a good idea for them to read their medical record. Another 34.1% agreed that their physician would think it was a good idea, 51% were uncertain, and 3.5% disagreed or strongly disagreed.
Patients were divided about their preferred mode of access. While half (49.3%) were interested in reading a paper copy of their medical record at home, a comparable number (43.8%) preferred obtaining access by an electronic version at a secure, private Web site. Written comments revealed the polarity of these views. One respondent wrote, "I do everything online. It would be great if all my records were accessible online and I could e-mail my doctor and/or nurse about a condition I am having and what they recommend." In contrast, another patient commented, "I do not want my records EVER put on a Web page. Nothing is private or secure. If they are, I will sue you royal."
This study reports the results of the first survey of patients' interest in reading their medical record in the era of the health consumer. Patients' interest is widespread and is associated with many attitudes and behaviors. Two of our initial hypotheses, however, were not supported. Interest is not a function of health status or health care use, nor is it a function of education or income. Interest is, however, driven by a general concern about health independent of heath status, by an interest in health information, and by concerns about patient safety and having a less trusting relationship with their primary physician. These findings suggest that a theory of health consumerism predicts interest in the medical record better than does a theory of clinical relevance.51 The findings lend credence to the concerns voiced by some that patient interest would be motivated by lack of trust.17 They also support others' contentions that patients' reading their medical records reflects a desire to be more involved in their own care.52
People who have demonstrated an interest in health by consulting other information channels such as print media and the Internet are very interested in reading their own medical record. The relationship with Internet use is linear (results not shown). This interest is a logical extension of their involvement in health and represents an opportunity for patients to use the most direct information source available about themselves.
Patients who have looked at their medical record in the past remain interested in reading it. This finding is consistent with earlier findings that patients who were given an opportunity to look at their records sustain that interest.37 It suggests that examining their medical record could become a routine, if not frequent, activity.
The current environment has heightened patients' concerns about medical errors, and this concern is directly related to an interest in reading the medical record. To the extent that we expect patients to take an active role in reducing medical errors,6 we have yet to exploit systematically 1 potentially useful tool—their own medical record. Patients could become more familiar with their own medical history and would be able to recognize and correct errors in their record.
In this population, patients with a less trusting relationship with their primary physician were more interested in reading their medical record. They may think that there is information in the record that has been hidden from them; they may think that there are some inaccuracies; or they may simply not understand their physician and are confused about their medical condition. Before launching the survey, we conducted qualitative interviews with 6 patients. Those who had strong relationships with their primary physician were relatively uninterested in their medical record because they thought that their physician had already shared all relevant information with them.
Even in this relatively homogeneous patient population, race remains a predictor of interest in one's medical record. Race may be a surrogate for a well-founded suspicion of the health care system.53 One means to dispel patient suspicion and increase trust in the health care system may be to provide ready access to the medical records.
We were not surprised to find that women are more interested in their medical record compared with men. Because women frequently serve as the health care decision makers in families, they may be attracted to this additional source of information.54- 56
Even in this well-educated population, one quarter of respondents were unaware that they had the legal right to examine their own medical record. In fact, believing that they did not have the right was associated with stronger interest in seeing the medical record—perhaps a "forbidden fruit" phenomenon.
The generalizability of the findings from this study is limited by the nature of the sample: suburban, middle-class Midwesterners, who were patients at 1 large multispecialty clinic. Nonetheless, it establishes a point of comparison against which the results of broader studies can be compared. The characteristics of this sample are typical of patients whom we expect to become most active in the burgeoning consumer movement and may represent the forefront of the demand for access to medical records. The study sample is also composed only of users of the health care system. People without access may exhibit other relationships with interest in reading the medical record.
The excellent response to the questionnaire (81%) lessens concern about response bias. It may also reflect a widespread interest in the topics covered by the survey.
We may be failing to use a potentially powerful tool, the medical record, to involve patients in their health care. In this population, over one third of patients report being very interested in reading their clinic records. Many are cautious, however, and would prefer to see a paper copy of their records rather than an electronic version. By providing access to the medical record in either paper or electronic versions, we may improve patients' understanding about their role in their condition, and we may improve the patient-physician relationship.
Corresponding author and reprints: Jinnet B. Fowles, PhD, Health Research Center, Park Nicollet Institute, 3800 Park Nicollet Blvd, Minneapolis, MN 55416 (e-mail: email@example.com.
Accepted for publication May 13, 2003.
This study was supported by a grant from Arlene M. Carlson, Minnetonka, Minn.
This study was presented at Minnesota Health Services Research Conference; February 26, 2002; Minneapolis; and Minnesota VA Outcomes Research Center; December 19, 2002; Minneapolis.
We acknowledge Beverly Gray for managing the data collection process.