Initially the LANE study comprised two cohorts, i.e. EX2002 and EX2004, in order to disclose potential cohort effects on the results. The basis for choosing year 2002 for initiation of the study was twofold. First, there was an urgent need to start the investigation in 2002 due to the increasing frequency of stress [25
] and long-term sick leave [27
] among nurses. Owing to the prospective longitudinal study design, the participants had to be still in education, and therefore it was appropriate to carry out the first data collection for cohort EX2002 in the last year of their education. At the same time, a more detailed investigation of the study period was mandatory. In order to secure data from all three years of education, cohort EX2004 was also formed in year 2002, but here study participants were in their first year of education. Second, the concurrent changes within the educational system, where higher nursing education increased in size, dimension [65
], started in 2003, i.e. between the formation of the two cohorts. In this way, changes in learning conditions can be studied in relation to possible short- and long-term effects on nurses' educational, clinical and health outcomes. Finally, the last cohort, i.e. EX2006, was formed later, with the same interval as the first two, in order to serve as yet another control group, both for the first two LANE cohorts and for a similar study on teachers (The PATH study; Prospective Analysis of Teachers' Health). Both EX2006 and the PATH cohort were in their final year of education in 2006.
The time frame selected in the LANE study for investigation of common career pathways, life transitions and change in psychological health, covers a period from studies to practice. This is a strength when evaluating the possible influence of educational, working and individual factors on health development and the retention of graduates in the profession. The number of time points in each cohort was chosen to maximize coverage before and after the adjustment and transitional process from student to graduate nurse. Described from a transitional perspective, our assessments were made at both ends of the transition to professional practice, i.e., before and after "the confusing nowhere of in-betweenness", where change is inevitable and the individual will need to develop approaches for dealing with it in one way or another [66
]. The transition was understood to affect individuals differently, but to affect participants physically, mentally and socially [66
] during the initial year of transition [67
]. Recently, Duchscher showed that after about a year, nursing graduates entering professional practice felt accommodated, and that this first year involved both personal and professional qualities. Also, as stated earlier, studies on professional turnover show that some new nurses leave nursing within a few years of entering the profession [15
]. Due to the problem of time lags being too short, involving a risk of missing the phenomena entirely, a longer time lag was chosen, since the risk is then merely a matter of underestimation [68
]. Hence, after one year most new graduates were assumed to have got beyond this initial phase, and since Zapf and colleagues (1996) recommend time lags that are too long rather than too short, one year seemed a reasonable approximation of optimal measurement. The one-year interval was chosen based on the idea that participants should have somewhat adapted to their new situation as registered nurses, and that equal time lags were recommended [68
]. With regard to longitudinal assessments of career pathways, life transitions and health changes during the first years of working life, it is an advantage to have repeated assessments at approximately the same time every year, due to seasonal differences and changes. Specifically, all assessments after graduation have been performed between February and April each year. This time also coincides with the collection of official statistics (by Statistics Sweden) regarding higher education and employment rates of new graduates [58
]. Thus, LANE data can efficiently be compared with official population statistics. In sum, the selection of time points for data collection was mainly based on striking a balance between two issues: 1) the aim to cover a wide variety of research areas (e.g. individual conditions, educational structures, health trends, mental ill health and well-being, and contextual factors in healthcare), and 2) the provision of maximum opportunity to compare the cohorts and control for above-mentioned cohort effects.
The LANE study resembles other concurrent cohort studies on new graduates' transition to practice, such as the Australian e-cohort study (including 540 nursing students) [69
], in that it focuses both on retention and employment patterns, as well as on prevalence of musculoskeletal symptoms and work-based injuries. However, unlike the e-cohort study, the main focus of the LANE study is on mental disorders and psychological well-being. Similar to a national cohort study in England (n = 2784), the sampling frame included students representing all nursing programmes in the country [60
]. In addition, these two studies both follow new graduates for three years after graduation, but differ in that LANE also includes data collections during the respondents' years in higher education.
Strengths and limitations
When data collection and analyses are complete, the LANE study will add unique knowledge, since surprisingly few studies have actually collected information both during education and after entering nursing practice [71
]. Duchscher emphasises that, although several studies now focus on investigating the effect of different orientation programmes on new graduates' experiences of moving into a professional nursing practice role, she has identified a lack of studies exploring pre-graduate transition preparation. Students' lack of familiarity with what awaits them after graduation, i.e. "the element of surprise", may have a negative effect on new graduates' professional role adaptation [71
Even if the wave response rates are generally high, they decrease over time. The possible selection bias introduced by this phenomenon must be carefully scrutinized in relation to each particular research question. Specifically, we will compare and contrast attrition due to leaving the profession, embarking on specialist training or being on maternity leave. Although our findings were not constant over the three cohorts, the analyses in this paper generally indicated that gender and country of birth influenced participation and retention. Firstly, the lower male participation (2% units) in the EX2006 cohort, as compared with the population is a phenomenon that has been reported earlier in similar studies. For instance, in the European NEXT (Nurses' Early Exit) study, there was a smaller proportion of men in the study sample, as compared with the percentage in the national workforce, in eight out of eleven countries [14
]. Similarly, 9% of the Australian nurse workforce was male, whereas only 6% participated in the study by Turner and co-workers [69
]. In the EX2002 cohort, men's response rate instead declined over time. This also seems to be a common trend in comparable studies, where fewer men than women responded to follow-up questionnaires [73
]. One possible explanation for the smaller number of male respondents could be linked to findings showing that men differ from women in that they: less often enter nursing as a first choice [77
], less often complete their education [56
], have a more critical view towards nursing education [58
] and are more inclined to leave the profession [79
]. As a result, men can be assumed to be less interested in participating in a study directly addressing nursing issues.
The number of immigrants consenting to participate in the EX2002 (6%) and the EX2006 (9%) cohorts was an underrepresentation as compared with the wider population (9% and 11% respectively). In this study, language difficulties cannot be ruled out as a reason for immigrant non-participation; however, this seems to be a less probable cause, since all participants were recruited from higher education. It is difficult to know whether cultural differences can explain the fact that immigrants were less likely to consent to participate, and whether respondents were more integrated into Swedish society than non-participants. In the EX2004 cohort, on the other hand, where the non-Swedish-born subgroup had lower response rates across time, this may be related to higher mobility and a tendency to move out of the country.
When understanding and interpreting different outcome areas in the LANE study, it is important to remember that the population of students at two sites of learning could not be defined prior to the study, and that students from these universities had to personally take the initiative to become part of the sampling frame. Although this selection may not be a major problem (at least not for the EX2002 cohort), exclusion of the consenting students from these two universities will be optional when comparing educational outcomes across universities with regard to data from the EX2002 and EX2004 cohorts. This reservation does not concern the EX2006 cohort, where the total population of students attending the last semester of the nursing programme in the autumn of 2006 could be defined in advance and included in the sampling frame. This limitation will be controlled for when contrasting educational data among the universities.
The main weakness of the study is that data are only collected through self-reports; thus, health data are not clinically validated. However, when data collection closes in 2010, additional data from national registers available for research will be used to form parallel cohorts; data on graduation, employment, maternity leave and sick leave will then be extracted and compared with LANE data.