Objective: To examine newspaper coverage of maternal health in three countries that have made varying progress towards Millennium Development Goal 5 (MDG 5): Bangladesh (on track), Rwanda (making progress, but not on track) and South Africa (no progress). Design: We analysed each country's leading national English-language newspaper: Bangladesh's The Daily Star, Rwanda's The New Times/The Sunday Times, and South Africa's Sunday Times/The Times. We quantified the number of maternal health articles published from 1 January 2008 to 31 March 2013. We conducted a content analysis of subset of 190 articles published from 1 October 2010 to 31 March 2013. Results: Bangladesh's The Daily Star published 579 articles related to maternal health from 1 January 2008 to 31 March 2013, compared to 342 in Rwanda's The New Times/The Sunday Times and 253 in South Africa's Sunday Times/The Times over the same time period. The Daily Star had the highest proportion of stories advocating for or raising awareness of maternal health. Most maternal health articles in The Daily Star (83%) and The New Times/The Sunday Times (69%) used a ‘human-rights’ or ‘policy-based’ frame compared to 41% of articles from Sunday Times/The Times. Conclusions: In the three countries included in this study, which are on different trajectories towards MDG 5, there were differences in the frequency, tone and content of their newspaper coverage of maternal health. However, no causal conclusions can be drawn about this association between progress on MDG 5 and the amount and type of media coverage of maternal health.
Total articles published for Bangladesh Rwanda and South Africa
We analyzed each country’s leading national English-language newspaper: Bangladesh's The Daily Star, Rwanda's The New Times/The Sunday Times, and South Africa's Sunday Times/The Times. We searched the electronic archives of each country’s newspapers described above from January 1, 2008 to March 31, 2013. We chose a January 1, 2008 start date because the searchable online archive of South Africa’s Sunday Times only goes as far back as January 1, 2008. We searched for articles on maternal health (including family planning) using the strategy detailed in Appendix A. The collected articles were entered on an Excel spreadsheet.
Coded articles Bangladesh
For the content analysis, we used a “constructed week” sampling technique, in which sample dates are stratified by day of the week to account for systematic variation due to day of the week. Through this “constructed week” sampling method, we sampled 75 articles from The Daily Star (Bangladesh).
Coded articles Rwanda
For the content analysis, we used a “constructed week” sampling technique, in which sample dates are stratified by day of the week to account for systematic variation due to day of the week. Through this “constructed week” sampling method, we sampled 64 articles from Rwanda’s The New Times/The Sunday Times .
Coded articles South Africa
For the content analysis, we used a “constructed week” sampling technique, in which sample dates are stratified by day of the week to account for systematic variation due to day of the week. Through this “constructed week” sampling method, we sampled 51 articles from South Africa’s Sunday Times/The Times.
Sampled articles to code
For the content analysis, we used a "constructed week" sampling technique, in which sample dates are stratified by day of the week to account for systematic variation due to day of the week. Through this “constructed week” sampling method, we sampled 75 articles from The Daily Star (Bangladesh), 64 articles from Rwanda’s The New Times/The Sunday Times and 51 articles from South Africa’s Sunday Times/The Times— a total of 190 articles published from October 1, 2010 to March 31, 2013
Inter-rater Reliability
Two coders (Frey Gugsa and Ellora Karmarkar) developed the codebook from relevant existing work, supplemented with additional themes that emerged from the three newspapers (Appendix B). We followed an iterative process of reading randomly selected articles, assessing themes, and adjusting the codebook until we achieved inter-rater reliability (Cohen’s Kappa > 0.82). All articles were recoded using the finalized codebook.