This study shows low participation in follow-up screening after gestational diabetes that are recommended in international and national guidelines
[5, 6]. The study of 2171 women with first-time gestational diabetes showed significantly decreasing participation in follow-up screening, from 80.5% at Control 1 to 17.7% of the women visiting their general practitioner for Control 4. Even lower attendance was found for biochemical departments, with 10.2% at Control 1 and 0.5% attending Control 4. The risk of a diabetes diagnosis was strongly associated with participation in follow-up screening, with an 11.8-fold increased risk for women attending general practitioner Control 4. Attenders at Control 4 had an increased risk of initiation of treatment with glucose-lowering agents at 8.9-fold and 8.7-fold for general practitioner and biochemical department controls, respectively. Women who attended at least one general practitioner control moreover had a 2.7-fold increased risk of shorter time to diabetes diagnosis. This also applied for the initiation of treatment, with a 2.1-fold increase in the risk for either type of control. The high risk of diabetes diagnosis and initiation of treatment among women attending the controls emphasizes that attendance enables early detection of diabetes, which can prevent the development of late complications
Strengths and limitations
The strength of this study is its basis in the complete and comprehensive data held by the Danish national registers and the negligible risk of loss to follow-up
[14, 26]. The structure of the Danish health care system and the possibility to include both hospital and general practitioners, as well as both register based diagnosis and initiated treatment allows a full follow-up on a large cohort. The defined intervals were furthermore sufficiently broad to ensure the detection of women who participated in follow-up screenings, whether at their general practitioner or at biochemical departments. The limitations of the study relates to the validity problems in connection with the regionally based blood sample registration, since this serves purely administrative purposes. We are unaware of any validity studies of on the regional blood sample registration. Our results on the proportion of women having tests performed at Control 1 are considerably higher than those found in comparable studies
[7, 8, 11, 27]. However, data on blood sampling have only been available since 2006, which has limited the analysis of testing at biochemical departments to women giving birth after that time. The insignificant results obtained for diagnosis of diabetes as outcome may be caused by the limited data. The problem of under-registration of diabetes diagnosis was caused by the fact that not all women in treatment with glucose-lowering agents had a registered diabetes diagnosis. Inaccuracies in coding procedures may have led to the problematic validity of diabetes diagnosis registered in the National Patient Register
. However, the National Patient Register enabled detailed information regarding hospital admissions in Denmark and further description of the register can be obtained from Lynge et. al.
. The detailed information of the National Prescription Registry, however, enabled us to detect women in diabetes treatment who were not registered with a diabetes diagnosis. Further information regarding the National Prescription registry can be obtained from Kildemoes et. al.
. As described earlier, type 2 diabetes is the more frequent diabetes type among women with previous gestational diabetes
. We expected that the majority of the diagnosed women had received a type 2 diabetes diagnosis. The diabetes outcome category contains all diagnosis of diabetes, which strengthens the validity of our study and prevented us from excluding women diagnosed with diabetes after gestational diabetes. The use of the National Health Service Register may also have caused validity problems, although we are unaware of any relevant studies regarding this matter. Further description regarding this register can be obtained from Andersen et. al.
. Despite the strong economic incentives, the misreporting of minor additional laboratory services in particular, may occur since fees for these services are small
. Whether the general practitioner performed oral glucose tolerance tests or blood glucose tests cannot be ascertained, as the codes used to detect follow-up screening are the same
[18, 19]. General practitioners who are unable to perform or analyse oral glucose tolerance tests can refer the women or have the test performed by a biochemical department. However, 80.5% of the women were registered as having a test performed at Control 1, which indicates a predominantly correct coding. A further limitation of this study concerns the possible demographic selection bias related to its basis in the relatively deprived North Denmark Region
[12, 29]. A higher proportion of socioeconomically disadvantaged women in our population may have led to an overestimated risk of diabetes related outcomes in relation to non-attendance in follow-up screening, since socioeconomic status is a determinant for diabetes
. Results regarding the risk of diabetes can be higher in our study population, and the relatively high risks have to be interpreted with caution. However a previous mentioned meta-analysis finds the risk of diabetes among women with pregnancy complicated by gestational diabetes, 7-fold higher compared to women with normoglycaemic pregnancies
. Only at the later controls our risk estimates exceeds the risk found in the meta-analysis.
While we found high participation rates in the first control after gestational diabetes, other studies have reported rates varying from 14 to 61% for this control. A recent review also finds that screening for diabetes after gestational diabetes generally is low
. However, those studies were restricted to the first control
. Organisational differences between health care sectors effects the comparability with results from other health care contexts. Attendance varied greatly for testing at first control, with 80.5% and 10.2% for general practitioners and biochemical departments, respectively. The majority of results retained significant levels, however the limited data on blood samples may have caused insignificant associations between controls at biochemical departments and diagnosis of diabetes and treatment at Control 3. A likely reason for the high participation in the first general practitioner control after giving birth is that this examination is part of a standard postpartum visit
. General practitioner controls had higher attendance than biochemical department controls, possibly because some practitioners are capable of performing the oral glucose tolerance test. However, other regions recommend that the tests are analysed or performed only by biochemical departments because of the varying quality of the general practitioners equipment
. Low participation in follow-up screening can be related to healthcare providers, health systems and patient barriers. Barriers related to healthcare systems concerns different practice between countries. Healthcare providers can be unaware of the guidelines and lack of communication between healthcare providers. General practitioners can also fail to perform the test or refer the women to biochemical departments
[8–10, 27]. The patient related barriers are lack of time and concerns about future health
. Women can also be unaware of the risk of diabetes after gestational diabetes
[8–10, 27]. The fragmentation of care may also account for low participation since this often leads to confusion about the responsibility for follow-up screening
[8–10, 27, 33]. Participation gives women a higher risk of diabetes diagnosis or treatment and an increased possibility of early detection of diabetes or initiation of treatment. The risk estimates were high at the last controls possibly due to the attendance of women who had been diagnosed or treated at earlier controls, although their attendance also decreased over time. The women who were not followed up were more likely to be diagnosed at a later point, when their diabetes may have progressed, since clinical diabetes can remain undiagnosed for years
. The high risk of a diabetes diagnosis among women attending controls and estimated undiagnosed diabetes among non-attending women, lead us to believe that a considerable number of the non-attenders can be affected by diabetes after gestational diabetes. We found the diabetes risk to increase with successive controls, which is in accordance with findings from other studies that show an increasing risk of diabetes within the first five years after gestational diabetes
. Women with high income had a significantly lower risk of diabetes diagnosis or treatment. Socioeconomic factors are plausible reasons for this, since they are one of the determinants for diabetes
. In view of the limitations described in previous sections, our results should be interpreted with caution. The main result of this study is the finding of low participation in follow-up screening after gestational diabetes, except for the first control after giving birth. Our study has demonstrated the importance of follow-up screening of women to ensure early detection and initiation of treatment for the prevention of late complications caused by diabetes. Follow-up screening of these women is a concern for public health as low participation impedes the detection of diabetes in women with previous gestational diabetes. This has been emphasized by a previous Danish study, which found that two out of three diabetes cases in the general population were undiagnosed
. The low participation in follow-up screening after gestational diabetes may reflect an ignorance of the risk involved, or failure by health staff to refer the patient and unawareness of guidelines
[8–10]. Fragmented care is another possible factor since the care of women is organised between obstetricians and general practitioners. The fragmentation can contribute to lack of communication regarding the recommendations for follow-up screening and the women’s future risk between the providers
[8–10, 33]. The study provides clinicians with important knowledge of the risk of diabetes among non-attending women and the need for communicating the risk to the women in order to ensure follow-up screening. In low-income countries, women with pregnancy complicated by gestational diabetes face greater barriers such as availability, affordability and access to services, which concerns both screening during pregnancy and access to follow-up screening
What this study adds
In contrast to other studies examining participation in follow-up screening
[7, 8, 11, 27, 35], this study followed women with previous gestational diabetes over several screening rounds and examined the consequences of non-attendance in follow-up screening. We document rapidly declining participation in follow-up screening with time and estimated that women not attending potentially can have undiagnosed diabetes. This study also shows that follow-up screening enables early detection of diabetes. Our study furthermore demonstrates the great potential of monitoring the quality of health care sectors through national registers.
Suggestions for further research
Our study has documented the need for preventive health interventions to ensure stringent follow-up screening of women with prior gestational diabetes in order to ensure the early detection of diabetes and initiation of treatment. The general lack of interventions to ensure participation in follow-up screening is described in other studies
[8, 27, 33, 35]. Our findings may furthermore help in the development of interventions to enhance participation in follow-up screening. Possible interventions could concern reduction of fragmented care and facilitation of other efforts to improve participation in follow-up screening. Such efforts could concern reminder systems, which has shown to improve the participation in follow-up screening significantly
[8, 11, 35, 36]. Such systems can have a positive impact on the participation in follow-up screening