Using injuries associated with three-wheeled all-terrain vehicles in Alaska as an example, the existing injury data bases were assessed for usefulness, cost, simplicity, acceptability, flexibility, sensitivity, specificity, representativeness, and timeliness. In this study strengths and weaknesses of existing data for all-terrain vehicles were identified and ways to improve data collection and linkages across data systems are suggested. Based on this evaluation, linked death certificates and medical examiner data provide an excellent mechanism for monitoring vehicle-related fatalities. Information sources for nonfatal and nonvehicle-related injuries require further development. Police records provide supplemental information, but they are limited to the events reported to police. Although other sources were explored, they added no advantage to the primary sources. Data processing, analysis, and dissemination--traditional responsibilities for public health and other governmental agencies--can transform these data sources into meaningful mechanisms to define injury trends and monitor injury-specific intervention strategies.
To evaluate the cost-effectiveness of a lottery on physicians' responses to a mail survey, a randomized controlled trial was conducted with a random sample of 1,000 members of the Quebec Federation of General Practitioners in 1997. For the first mailing of this survey, each respondent was randomly assigned to the control or experimental group, which was offered participation in a lottery upon return of the questionnaire. Response rate was 41.2% in the experimental group and 34.8% in the control group, a 6.4% difference (CI95%: 0.6%-12.6%). The additional cost of the lottery was about Can$500, giving an incremental cost of Can$16 per questionnaire returned. In conclusion, a lottery resulted in a small but statistically significant increase in the response rate of physicians to a mail survey. This method may be a cost-effective option when applied to large surveys.
Most severe diseases cannot be prevented. The main reason is lack of information about their causes. The Norwegian Mother and Child Cohort Study (MoBa) is a population-based cohort with 100,000 children included from fetal life. The ambitious goal is to understand the causal chains behind diseases and the comprehensive data collection will allow many diseases to be addressed. The overall goal is to come up with good preventive measures. The purpose of this article is to inform about MoBa and to stimulate enhanced use of data collected during research.
Public Health agencies need up-to-date population data, and Statistics Canada's Taxation Family File (T1FF) database, developed annually from T1 personal income tax forms, may meet this need. We compared 1989 T1FF data for Ottawa-Carleton with census data. The overall population estimate for the region was 98% of the census figure. Equivalent figures were 94% for central cities, 98% for suburbs and 122% for rural areas. The T1FF underestimated the population of children (89%) and seniors (93%). Population estimates for smaller neighbourhoods were poorer, some being wrong by a factor of more than two. The errors are probably due to postal code conversion. The T1FF provides fairly accurate estimates of the population of larger areas, and can be much more current than census data. It is less accurate for smaller areas, offers a limited range of variables, does not specify gender for children, and is fairly expensive. Health units with a major commitment to needs-based planning should consider using the T1FF for denominator data, despite its limitations.
To compare two ways of combining postal and electronic data collection for a maternity services user-experience survey.
Maternity services in Norway.
All women who gave birth at a university hospital in Norway between 1 June and 27 July 2010.
Patients were randomized into the following groups (n= 752): Group A, who were posted questionnaires with both electronic and paper response options for both the initial and reminder postal requests; and Group B, who were posted questionnaires with an electronic response option for the initial request, and both electronic and paper response options for the reminder postal request.
Response rate, the amount of difference in background variables between respondents and non-respondents, main study results and estimated cost-effectiveness.
The final response rate was significantly higher in Group A (51.9%) than Group B (41.1%). None of the background variables differed significantly between the respondents and non-respondents in Group A, while two variables differed significantly between the respondents and non-respondents in Group B. None of the 11 user-experience scales differed significantly between Groups A and B. The estimated costs per response for the forthcoming national survey was ?11.7 for data collection Model A and ?9.0 for Model B.
The model with electronic-only response option in the first request had lowest response rate. However, this model performed equal to the other model on non-response bias and better on estimated cost-effectiveness, and is the better of the two models in large-scale user experiences surveys with maternity services.