Survey responses between the two groups consistently indicated no statistically significant differences concerning participant perceptions and reflections on the mentoring program. Similarly, survey responses taken over time also failed to detect the presence of statistically significant temporal trends, be they increasing or decreasing.
Face-to-face meetings were found to be the most popular type of interaction, similar to findings recently reported by Raghunanda et al. [28]. As a second preference, email and telephone communications were of similar frequencies. Also, these interactions, irrespective of type, were not frequently undertaken during the program’s running. The majority of pairings interacted at most four times. This lack of frequency could explain why mentee and mentor perceptions and recollections (e.g. topics discussed during interactions) were not always in total sync. de Janasz & Godshalk,1 have postulated that a lack of interaction frequency could be one of the retarding factors in assisting mentees to ‘mimic’ mentor behaviours and maximise benefits.
Both groups reported high levels of satisfaction with the program per se and the degree to which it impacted the satisfaction of training and/or working in the CCLHD. Though satisfaction scores were consistently high for both questions, mentees were slightly elevated on the former and reversed on the latter. These results very much reflect past findings that mentoring is not mutually exclusive, and more is likely to benefit both parties involved in the mentoring process – not just the mentee [29–32].
One recent systematic review examined the prevalence of mentoring in medicine and its effect on career development [24]. They found a significant relationship between the two. Though this relationship was not explicitly quantified in our study, there was sufficient study feedback to indicate that the topic of career development was of extreme importance to our program participants. Along with life/work balance, it was the one topic that was discussed more frequently than any other topic or issue during mentee-mentor interactions.
The main strengths of this study were that it was both longitudinal and quantitative, unlike most studies that have focussed on mentoring programs that have been cross-sectional and/or qualitative [10, 33–35]. As such, it was able to explore and quantify mentee/mentor perceptions and reflections over time. It also had the advantage of being a multi-site study instead of a single site study, thus allowing its findings to be more generalisable than if participants had been sourced from either the GDF or the WPH, but not both. Lastly, but most importantly, mentors were allowed to receive training and resources before the program commenced. A lack of appropriate training can lead to the delivery of inappropriate advice or the imposition of the mentor’s views on the mentee, resulting in conflict and disillusion [31, 36].
There are several caveats to our findings. Firstly, surveys were done via an online questionnaire, thus generating concerns regarding potential selection and reporting biases [37, 38]. That being said, the response rate was relatively high for both mentors and mentees and at all time-points. Secondly, and due to the small sample size, it may be well that the study was underpowered, thus producing type II errors and non-significant test results. Thirdly, the quantification of program effects could have been enhanced if more survey items had allowed ordinal scale responses rather than dichotomous responses. It has been shown that reliance on the latter leads to a loss of information and effect size and power [39].
Although the impact and importance of mentorship have been discussed at length, tangible suggestions for improving mentoring programs are equally valuable. To address these limitations, it is essential that the following be considered if this program is to be extended in its current settings or expanded beyond the confines of the CCLHD. Firstly, findings from future mentoring programs would be more generalisable and statistical analyses would be more powerful if the number of participants was increased. This could be achieved by increasing the response rate through reminders and incentives and/or expanding the program to include one or more adjacent health districts. To this end, it would also be beneficial if eligible non-participants were identified, such as junior doctors deemed to be in most need of a mentee-mentor relationship, plus mentors who are genuinely vested in the mentoring process. One only to peruse the ‘mentoring landscape’ to realise, that comparable mentoring programs in a medical setting have been conducted (and are being conducted) with participant numbers far in excess of those achieved in this study.40 Secondly, gender imbalance in mentor participation needs to be addressed. While mentors of either gender should be equally effective, especially for female members, to support their academic activities [41–43], female mentors might be necessary for female mentees for a various reasons, such as their ability to serve as role models on combining the demands of their career with family commitments and better understand female mentees undergoing training [42, 44, 45]. Therefore, it is necessary to cultivate and recruit female senior medical staff as mentors. Thirdly, if the chosen survey was both valid and reliable, the quality of evidence from future studies on mentoring programs would be strengthened immensely.