Evaluation of An Online Intervention In Improving General Practitioners' Practice In Prostate Cancer Screening (ProCaSS)

Study design: Randomised controlled trial

Setting: This trial will involve GPs in the Klang Valley, encompassing the Petaling district and Kuala Lumpur.

The healthcare system in Malaysia is divided into public and private sectors. Patients pay a standard minimal fee for public healthcare system whereas the private sector charges patients based on the services provided. The study will be conducted in private GP clinics as PSA testing is readily available compared to the public primary care setting. In public primary care clinics, PSA screening is not offered as part of a routine screening programme.

Sample size:

With 80% power and 5% (two-sided) significance, with an estimated 40% reduction in PSA screening from baseline in the intervention group and 10% in the control group, the number of participants needed for each group is 38 (total 76). Therefore, a total of 96 GPs in the Klang Valley will be recruited for this study, assuming a non-response rate of 20 GPs.

Intervention:

The intervention consists of two phases.

Phase 1 The research team will develop 3 self-administered, online surveys, with the input of family physicians and urologists. The surveys will capture basic demographic information about the participating GPs (but not any identifiable information), and also their overall knowledge and perception towards prostate cancer screening. In addition, all 3 surveys will also contain 5 clinical vignettes that will relate to prostate cancer screening. After reading the vignettes, the GPs will be asked whether or not they would recommend screening for the particular patient in the given scenario; and should they choose to screen, which screening method they would use (PSA, DRE, or both).

GPs will then be randomised to either receive an online training module (intervention) or none (control).

Phase 2:

Upon successful receipt of the GPs' response to the first survey, a second survey will be sent to all the GPs via email. The participants will once again be invited to read and respond to another set of vignettes of similar nature to those in phase 1.

GPs who have been randomised to the intervention arm will also receive a link in the second email that will enable them to download an online training module. The participants will be asked to view the training module prior to answering the second survey.

GPs in the control group will only receive the second survey and not the training module.

A third and final email will be sent to all the GPs who have successfully answered the second survey, 3 months after the date of receipt of their second survey response. This final survey will likewise, contain five clinical vignettes of similar nature to those of the two previous surveys. The responses of GPs in the intervention group will then be compared before and immediately after the online training video and also 3-months later; as well as to the control group.

Recruitment:

The research team will recruit the GPs from an existing GP database which has been created from previous studies conducted in the Petaling District and Kuala Lumpur. A researcher and a research assistant will be responsible for identifying and recruiting eligible participants. An invitation letter, a study summary, a participant information sheet and consent form will be sent via email to each GP. Consenting participants will be remunerated for their time and effort spent in participating in the study. In the event response is poor, a research assistant may need to arrange an appointment to visit the GPs at their clinics'.

Analysis:

The McNemar test will be used to determine the phase differences in the proportion of GPs who perform screening unnecessarily and the differences between the intervention and control group. Chi square test will be used to compare the difference in proportions between the intervention and control groups. GPs' characteristics associated with unnecessary screening practice will be explored using logistic regression models using phase 1 data. Logistic regression will be used for binary outcomes and the analyses will be adjusted for baseline data. SPSS will be used to manage the data sets