Skip to main content

Table 2 Summary of the results of data accuracy assessed in six randomized controlled trials

From: A review of randomized controlled trials comparing the effectiveness of hand held computers with paper methods for data collection

Study Results related to data Accuracy Definitions
  Hand Held Computers Paper  
Tiplady B et al. 1997 38 Missing data 8.91%
Problematic data** 5.64%
Missing data 0.16%
Problematic data** 0.24 %
Accuracy was defined by comparing missing and problematic data between the two methods.
**Problematic data points defined as those requiring some intervention or editing.
McBride JS et al. 1999 36 No difference in missing item responses between PDA and paper in 4/5* subscales, (p < 0.05). No differences in missing item responses between PDA and paper in 4/5* subscales, (p < 0.05). Defined as a comparison of missing item responses between the two methods.
Lal SO et al. 2000 35 2.8% error frequency. 6.7% error frequency Data fields analyzed for frequency of error were gender, race, date of birth, date of burn, date of admittance to hospital, and burn type. Accuracy determined by comparing these fields with original medical record.
Jamison RN et al. 2002 34 Of 503 paired verbal stimuli in 24 subjects the correlation between paper and PDA ratings was r = .97 (range 0.95–0.98), for sensory stimuli r = 0.86 (range 0.81–0.92). Correlation between group electronic VAS and paper VAS ratings to common verbal stimuli r 2 = 0.997, for the common sensory stimuli group correlation was r 2 = 0.99. Defined as the degree of correlation between the two methods of rating.
Quinn P et al. 2003 33 Errors not possible in electronic diaries due to prompt and format of questions and responses. Errors "detected" in 80% of paper diaries Errors defined as incomplete times, inconsistent timing of events, incomplete events, incorrect completion of VAS.
Walker I et al. 2004 32 3 vials/patient not accounted for.
15 patients with errors.
5 vials/patient not accounted for (P = 0.45).
13 patients with errors (P = 1.00).
Both the number of vials/patient not accounted for in each group and the number of patients in each group with errors.
  1. *The subscale in which differences in missing item responses were found was the 1st response choice on PDA, authors suggest this could be attributed to learning effect.