Comprehensive overview of computer-based health information tailoring: a systematic scoping review

Abstract

Objectives: To explore the scope of the published literature on computer-tailoring, considering both the development and the evaluation aspects, with the aim of identifying and categorising main approaches and detecting research gaps, tendencies and trends.

Setting: Original researches from any country and healthcare setting.

Participants: Patients or health consumers with any health condition regardless of their specific characteristics.

Method: A systematic scoping review was undertaken based on the York's five-stage framework outlined by Arksey and O'Malley. Five leading databases were searched: PubMed, Scopus, Science Direct, EBSCO and IEEE for articles published between 1990 and 2017. Tailoring concept was investigated for three aspects: system design, information delivery and evaluation. Both quantitative (ie, frequencies) and qualitative (ie, theme analysis) methods have been used to synthesis the data.

Results: After reviewing 1320 studies, 360 articles were identified for inclusion. Two main routes were identified in tailoring literature including public health research (64%) and computer science research (17%). The most common facets used for tailoring were sociodemographic (73 %), target behaviour status (59%) and psycho-behavioural determinants (56%), respectively. The analysis showed that only 13% of the studies described the tailoring algorithm they used, from which two approaches revealed: information retrieval (12%) and natural language generation (1%). The systematic mapping of the delivery channel indicated that nearly half of the articles used the web (57%) to deliver the tailored information; printout (19%) and email (10%) came next. Analysis of the evaluation approaches showed that nearly half of the articles (53%) used an outcome-based approach, 44% used process evaluation and 3% assessed cost-effectiveness.

Conclusions: This scoping review can inform researchers to identify the methodological approaches of computer tailoring. Improvements in reporting and conduct are imperative. Further research on tailoring methodology is warranted, and in particular, there is a need for a guideline to standardise reporting.

Keywords: health informatics; information technology; public health.

Figure 1

The flow of information through the different phases of the review.

Figure 2

Classification map; an overview of the identified categories in each aspect. CD, compact disc; NLG, natural language generation; UX, user experience.

Figure 3

The frequency of articles in each of the health domains categorised by year of publication.

Figure 4

Co-occurrence matrix between categories of user profile. The abbreviations that are used in the table includes: UI, user interactions); UP, user perception; H/M, health/medical; psych, psycho-behavioural determinant; SD, sociodemographic; TB, target behaviour.

Figure 5

The frequency of delivery channels represented as the word cloud.

Figure 6

Co-occurrence matrix between categories of evaluation indicators. H/M, health/medical; psych, psycho-behavioural determinant.