Does the Use of an Automated Tool for Self-Reporting Mood by Patients With Bipolar Disorder Bias the Collected Data?

Michael Bauer, MD, PhD; Natalie Rasgon, MD, PhD; Paul Grof, MD, PhD; Laszlo Gyulai, MD; Tasha Glenn; Peter C. Whybrow, MD

In This Article

Abstract and Introduction


Context: Automating data collection from patients can improve data quality, enhance compliance, and decrease costs in longitudinal studies. About half of all households in industrialized countries now have a home computer.
Objective: While we previously validated the ChronoRecord software for self-reporting mood on a home computer with patients who have bipolar disorder, this study further investigates whether this technology created a bias in the collected data.
Methods: During the validation study, 80 of 96 (83%) patients returned 8662 days of data (mean, 114.7 ± 32.3 SD days). The patients' demographics were compared with those of similar longitudinal studies in which patients used paper-based data collection tools. In addition, because demographic characteristics may influence attitudes toward technology, observer-rated scores on the Hamilton Depression Rating Scale and Young Mania Rating Scale were used to group patients by severity of illness, and the self-reported mood ratings were analyzed for evidence of bias from the patients' gender, ethnicity, diagnosis, age, disability status, or years of education. Analysis was performed using the 2-way analysis of variance and general linear model.
Results: The patients' demographic characteristics were very similar to those of patients with bipolar disorder who participated in comparable longitudinal studies using paper-based tools. After grouping the patients by severity of illness, none of the demographic variables had a significant effect on the patients' self-reported mood using the automated tool.
Conclusion: The use of a computer does not seem to bias sample data. As with studies using paper-based self-reporting, results from studies of patients using ChronoRecord software on a home computer to report mood can be generalized.


Bipolar disorder is a leading cause of disability among both physical and psychiatric illnesses[1] and is the most expensive psychiatric diagnosis in the United States for patients and their insurance plans.[2] Bipolar disorder is difficult to study, as the course of disease is episodic, recurrent,[3] and characterized by both interindividual variation and heterogeneity among patients.[4,5,6] While the inherent complexity and chronicity make longitudinal studies an effective methodology, there are several problems with the paper-based instruments most common in these studies: the National Institute of Mental Health (NIMH) LifeChart Method[7]; the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) Mood Chart[8]; and the ChronoSheet.[9] Specifically:

  • Data entry from paper-based forms is expensive, slow, and associated with high error rates that affect the quality of the data.

  • Patients frequently complete paper-based forms sporadically,[10] often just before a study visit, and retrospective recall can be inaccurate.[11]

  • Longitudinal studies are often associated with high rates of missing data and unbalanced numbers of observations from participants.[12]

  • Patients with conditions, such as chronic pain or asthma, often prefer an automated approach, thereby increasing compliance with electronic rather than paper diaries.[13,14]

To improve on the available paper-based tools, we developed an automated tool (ChronoRecord) for daily self-reporting of mood, sleep, and medications by patients with mood disorders.[15] In our validation study, 80 of 96 (83%) patients with bipolar disorder from 3 locations showed high acceptance of the computerized approach, entering 8662 days of data for a 3-month period (mean, 114.7 ± 32.3 SD days). The mean percentage of days missing for mood data was 6.1% ± 9.3 SD, which was equivalent to missing 7.3 of the 114.7 days. Concurrent validity was found between the observer ratings on the Hamilton Depression Rating Scale (HAMD)[16] and the self-reported ChronoRecord mood ratings.

While it is generally accepted that using a paper-based data collection tool does not yield biased data from patients with bipolar disorder, automating self-reporting of mood ratings may bias the data because: (1) only patients who are able to use a home computer are included; and (2) patients have varying levels of computer literacy. Several approaches were previously used to analyze data collected through an automated process for potential bias. Analysis of missing data showed no relation among illness severity or demographic characteristics -- data were missing completely at random.[15] No statistically significant difference was found between the demographics of the 80 patients in the study who returned data and the 16 who did not. This study further explores these 2 issues related to potential bias in data collected through use of a home computer.

To address the first issue, demographic analysis of the participants in the validation study was expanded for comparison with the demographic profiles of patients who used paper-based tools. To address the second issue, patients were grouped by severity of illness according to observer ratings on the HAMD and Young Mania Rating Scale (YMRS).[17] In addition, self-reported mood ratings on ChronoRecord were compared for each of the available demographic categories, since such characteristics as age,[18,19,20] gender,[21,22] ethnicity,[23] and disability[24,25] may be associated with different levels of familiarity and comfort with computer technology.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.