A cross-sectional paper-based survey was administered to all senior emergency physicians (site A: n=50; site B: n=39) in the EDs of 2 large public metropolitan teaching hospitals in Australia (Table 1). The sites were selected using convenience sampling. Senior emergency physicians were defined as department directors, specialists in emergency medicine, registrars, and senior medical officers and were included based on their involvement and responsibility in relation to test ordering and test result follow-up. The study was approved by the ethics committees from both hospital study sites and the University of New South Wales, Sydney, Australia.

A survey consisting of 22 closed-ended questions was developed based on literature related to patients’ and physicians’ access to clinical information [34,37]. We pilot-tested the survey face-to-face with 2 senior hospital specialists (1 ED physician and 1 rheumatology specialist) from Australia and 10 US primary care practitioners. Based on their feedback, modifications were made to improve clarity of the survey. Fifteen of the survey questions had 7-point Likert response scales with options ranging from strongly agree to strongly disagree.

A detailed description of the design and development of the survey instrument is included in the paper on primary care physicians’ attitudes to directly releasing results to patients [38]. An identical survey was used in both studies. Questions included: current practices and institutional policies for patient notification of abnormal test results (timing, methods, and responsibilities), attitudes toward direct notification of normal and abnormal results, possible physician concerns including lack of patient expertise in interpreting results, previous experiences with missed abnormal test results by themselves or others, patient care decision-making attitudes, and participant demographic details.

For the purposes of the study, direct notification was defined as the automatic release of test results directly to the patient regardless of whether or not the ordering physician had reviewed the results. Direct notification could be achieved by mail, telephone, fax, SMS text message, or by patients accessing their EMR using a patient portal. Abnormal test results were defined for this study as clinically significant abnormal results. This includes those that are not immediately life-threatening but required short-term follow-up; for example, newly elevated glucose blood levels or a chest x-ray with a new shadow.

The survey was self-administered by emergency physicians at the 2 study hospitals between July 1 and September 30, 2012. The survey took each physician approximately 15 minutes to complete. After obtaining staff rosters, researchers approached each ED physician from the study population and asked them to participate in the study. Each physician was informed of the confidential nature of the results and each survey contained a participant information sheet outlining the nature of the study and that completion of the survey implied consent. The survey was left with the ED physician and they were asked to place the completed survey in a dedicated secure box in the ED secretary’s office. The researchers returned at the end of each week to collect the completed surveys and follow-up with nonrespondents. Those physicians who had failed to complete the survey were reminded and given another survey if the previous one had been mislaid.

Data were analyzed using SPSS version 21.0 (IBM Corp, Armonk, NY, USA) and descriptive statistics calculated. For ease of interpretation, we recoded the dependent variable responses into dichotomous categories. For descriptive analyses and logistic regression, the responses of agree, moderately agree, or strongly agree and disagree, moderately disagree, and strongly disagree were collapsed into agree and disagree, respectively. The category of neither agree nor disagree was retained for descriptive analyses.

We used logistic regression to identify predictors related to emergency physicians’ comfort with direct notification of abnormal test results to patients and their agreement with direct notification of normal test results. Exact methods for logistic regression models were adopted because the sample size of the study was too small for asymptotic methods [39]. The category of neither agree nor disagree was excluded for logistic regression because the number of physicians who responded to this category was small and exact analysis for an ordinal logistic regression model was not available computationally. Two exact logistic regression models were developed using a stepwise forward selection of covariate method [39]. To avoid overfitting the models, we included no more than 2 explanatory variables in each model [40]. We included only variables that were clinically relevant as possible predictors. Predictors in the final models were considered significant at the P<.05 level. Odds ratios and their 95% confidence intervals were presented for the predictors included in the final models. The regression analysis was generated using SAS/STAT version 9.3 (SAS Institute, Inc, Cary, NC, USA).

A total of 89 emergency physicians were invited to complete the survey and 61 complied giving an overall response rate of 69% (61/89) (Table 2). The response rate for hospital A was 74% (37/50) and for hospital B was 62% (24/39).

Approximately half of the emergency physicians (54%, 33/61) were not comfortable with patients receiving direct notification of abnormal results, although a similar proportion (57%, 35/61) agreed with direct notification of normal results (Table 3). Most (57%, 35/61) agreed that overall a direct notification system would reduce the number of patients lost to follow-up. Physicians’ major concerns relating to direct notification of abnormal results were confusion (92%, 56/61), lack of expertise necessary to interpret test results (90%, 55/61), and patient anxiety (85%, 52/61). Most also expressed concerns that the patient may seek information that could be unreliable (57%, 35/61). Most physicians were not concerned that direct notification would interfere with the practice of medicine (90%, 55/61) or increase their workload (92%, 56/61). The majority of physicians disagreed with releasing abnormal results directly to patients for human immunodeficiency virus (HIV) (84%, 51/61), cancer screening (84%, 51/61), and chest x-rays (59%, 36/61). Tests which most thought suitable for direct notification were lipid profile (64%), blood glucose (59%), and urinalysis (57%). Approximately half the respondents felt comfortable with releasing abnormal complete blood counts (49%, 30/61), thyroid function tests (48%, 29/61), and electrolytes (44%, 27/61) directly to patients (Table 3).

Most respondents agreed (75%, 46/61) that there were standard policies and procedures for notification of abnormal results in their EDs (Table 4). However, there were mixed responses regarding who should be responsible for notifying patients of results with some physicians (43%, 26/60) stating that it was not always clear who should notify patients of abnormal results. The majority agreed that the ordering physician (or their assigned delegate) should be solely responsible for notifying patients (65%, 39/60) and the majority also agreed that the primary care provider should be responsible for following up abnormal results regardless of who ordered the test (65%, 39/60). Emergency physicians’ practices in relation to methods of patient notification of abnormal test results varied with respondents stating that they always (47%, 25/53) or sometimes (43%, 23/53) phoned the patient. The majority of emergency physicians indicated that they never emailed (100%, 44/44), never faxed (91%, 40/44), and never waited for the patient to contact them (91%, 40/44) regarding abnormal results. In relation to decision making, most respondents (58%, 35/60) thought that they shared responsibility with the patient for deciding on treatments.

In the previous 12 months, 21% (13/61) of emergency physicians said they had missed an abnormal result which led to delayed patient care, although half said they did not know whether they had missed a result or not (51%, 31/61). When asked if their colleagues had missed an abnormal result, a higher proportion responded yes (44%, 27/61). Participants reported radiology results to be the most frequently missed (38%, 23/61) followed by microbiology (13%, 8/61) and chemistry (13%, 8/61).

The final exact logistic regression model showed that emergency physicians who were not concerned about whether patients might seek unreliable information were 4 times more likely to be comfortable with direct notification of abnormal test results than those who were not concerned about this issue (OR 4.56, 95% CI 1.04-24.30). If direct notification became the norm, physicians would feel more comfortable with releasing abnormal test results on blood glucose than on other test results (OR 15.74, 95% CI 2.84-171.16).

Emergency physicians’ agreement on whether a direct notification system would reduce physician workload was the main factor associated with agreement on direct patient notification of normal test results. Those who agreed that a direct notification system would reduce physician workload were more likely to agree with a direct patient notification of normal results than those who disagreed that it would reduce workloads (OR 5.72, 95% CI 1.14-39.76). Those who were neutral about this issue were more likely to agree on direction notification than those who disagreed, but this difference was not statistically significant (OR 3.84, 95% CI 0.72-27.81).

Although most emergency physicians in our study thought that a direct test notification system would reduce the number of patients lost to follow-up, just over half did not support direct notification of abnormal test results. The main concerns expressed were that it would result in patient anxiety and confusion, and that patients lacked the necessary expertise to interpret test results. Physicians were more likely to be comfortable with direct notification of abnormal test results if they were not concerned about whether patients might seek potentially unreliable information. However, the majority of emergency physicians in our study were comfortable with direct notification of normal test results and they were more likely to be supportive if they thought this would reduce their workload. Approximately 1 in 5 respondents reported missing results in the previous 12 months, whereas half did not know whether they had or not. Radiology results were cited as the most frequently missed. There were diverse responses regarding who should be responsible for notifying patients of abnormal results with a number of respondents stating that this was not always clear.

Two key results from our study, which have not been previously published, focus attention on the unique ED environment, where physicians normally have no continuing relationship with their patients. Most respondents in our study notified patients of abnormal test results via telephone, sometimes they used a letter, and email was never used. ED physicians also showed mixed views regarding who was responsible for notifying the patient of an abnormal result. Lines of responsibility for test notification seem unclear because the majority of respondents in our study thought the physician who ordered the test should be responsible, but the majority also agreed that the primary care provider was responsible as well. These new findings underscore the need to design test notification systems to suit the ED context.

The data were collected from senior emergency physicians from 2 EDs. Emergency physicians from regional and smaller hospitals and junior emergency physicians may have different attitudes. We did not include a question in the survey about the possible responsibility of laboratories (pathology or imaging) with regard to notifying patients of test results, which meant we did not garner emergency physicians’ opinions on this. Qualitative interviews with emergency physicians, laboratory personnel, and information technology staff could have elicited richer information on barriers and facilitators to patients having direct access to their test results.

Future work needs to determine if direct notification of test results to patients leads to improved follow-up of abnormal results. Methods of ensuring patients can access test results directly, including secure Web-based patient portals with drill-down capabilities providing test descriptions and result interpretations need to be evaluated in terms of patient outcomes, cost, and patient usability across socioeconomic groups. Implications of patients “pulling” results from a laboratory information system using a patient portal tied to a hospital-based EMR versus test results being “pushed” automatically to patients needs evaluation.

The fast-paced ED environment presents a number of unique challenges for test result follow-up. Although notification of test results to ED patients has promise, it is important to ensure that methods for direct notification suit the environment and work practices of ED clinicians and laboratories and meet patient needs. Efforts should be directed toward establishing a clear set of recommendations regarding which test results should be directly notified to which patients, methods of notification, and the timing of notification. Emergency physicians’ concerns regarding anxiety, confusion, and lack of patients’ expertise to interpret results should be addressed in order to promote wider test results access to patients.