Increasing Screening for Type 2 Diabetes in Women With Previous Gestational Diabetes.

In 1998, the Canadian Diabetes Association (CDA) published guidelines that recommended all women diagnosed with gestational diabetes mellitus (GDM) should be screened 6 weeks to 6 months postpartum for type 2 diabetes mellitus (DM) using a 2-hour 75g oral glucose tolerance test. We determined the rate of screening for type 2 DM at the Ottawa Hospital before and after the publication of this guideline and found that none of the women in either period had been screened following the CDA guideline recommendations although other screening tests that had not been recommended had increased in utilization. Since the publication of the CDA guidelines several studies have been published that demonstrate that identification of patients with an abnormal oral glucose tolerance test and early treatment prevents the onset of diabetes. This research and the recommendations need to be communicated to physicians.

Methods:

Overall Objectives:

1. To test whether a reminder letter outlining the screening recommendations that is sent both to the patient and her family physician would increase screening postpartum for type 2 DM in women with previous GDM more than a letter sent only to the physician or the patient.
2. To determine the patient and physician barriers that are impeding the adoption of this recommended screening strategy.

Setting:

The High Risk Obstetrical Clinics (HRU) at the Ottawa Hospital Civic and General Campus.

Inclusion and Exclusion Criteria:

All women, who attend the HRU and who have been diagnosed with GDM will be considered for inclusion in the study. Women will be approached for participation in this study by the HRU health care team. If the patient agrees, then the research team will contact the patient to describe the study and obtain informed consent. (Appendix 1A)

Exclusion criteria:

1. Patient does not have a Family Physician
2. Patient's Family Physician already has a patient enrolled in the study
3. Patient already enrolled in the study for a previous pregnancy
4. Unable to give informed consent in English or French
5. History of pre-gestational diabetes mellitus
6. Intra-uterine fetal death or still birth

Study design and Randomization:

The design is a 2 X 2 factorial randomized controlled trial stratified by clinic location. Randomization will be performed by the Research Coordinator using a computer generated randomization number and a variable block size of 4 and 8. The HRU team, the investigators, and the patient will be blinded to the patient's group allocation. This process will decrease selection bias. Patients will be randomized postpartum by the research coordinator (to ensure that the patient does not have any of the exclusion criteria) to one of the following four groups.

1. Physician and patient reminder
2. Physician reminder with no patient reminder
3. No physician reminder but a patient reminder
4. Neither physician nor patient reminder. (Current standard of care)

Pre-Intervention Data Collection:

Baseline demographic data, maternal characteristics that have previously been shown to be associated with type 2 DM, and obstetric and reproductive history will be collected at the time of consent or at a maximum of two weeks after consent. The mailing address and telephone number of the patient and their family physician will also be obtained. A chart abstraction will be performed approximately 30 days after the expected date of confinement (EDC) to determine if a live birth occurred and the weight and gender of the baby will be abstracted. (Appendix 1B)

Intervention:

The screening recommendation strategy will be randomly allocated by the research coordinator after confirmation of a live birth and ensuring that the mother and her baby have been discharged from the hospital. In the unlikely event of a death, it was felt that the patient should be screened by the health care team with follow-up in the HRU postpartum as type 2 DM may be related to the cause of death. The physician reminder letter will consist of a patient specific recommendation to screen postpartum for type 2 DM using a 2-hour 75g OGTT and the evidence supporting this recommendation. The letter will inform the physician if their patient has or has not been given the requisition for the recommended screening test. (Appendix 1C) The patient reminder consists of a letter requesting the patient to proceed with screening, the requisition that is required for the screening test, as well as the instructions to follow for screening. (Appendix 1D)

Primary and Secondary Outcome:

Primary outcome:

The primary outcome will be the proportion of patients screened for type 2 DM using the recommended test, a 2-hour 75g OGTT. A Mantel-Hanszel chi-squared test statistic for overall comparison of the proportions in the 2 X 2 design will be used. As well Fisher's exact test for comparison of specific pair wise group proportions will be employed.

Ascertainment of the Primary Outcome

For patients who are randomized to the patient letter and requisition, the HRU health care team will receive the results of the test. For patients who are randomized to the physician reminder, the reminder will request that on the laboratory requisition that a copy of the results be forwarded to the HRU health care team. Also, for all patients screening will be ascertained by the following methods:

1. At the patient follow-up phone interview, the patient will be asked if they were screened and what test was used.
2. At the physician follow-up phone interview, the physician or nurse will be asked if the patient was screened, what test was used, and what were the results.

All patients will have a similar process and similar intensity in the follow-up of the screening results. We expect with this method of ascertaining the primary outcome we will be able to determine whether the patient was screened or not. The investigator who is gathering the information from the physician or patient will be blinded to the patient's group allocation.

Secondary Outcomes:

1. Proportion of Patients screened with tests other than the recommended test A Mantel-Hanszel chi-squared test statistic for overall comparison of the proportions in the 2 X 2 design will be used. As well Fisher's exact test for comparison of specific pair wise group proportions will be employed depending on whether or not an interaction effect was found.
2. Characteristics associated with screening: A univariate analysis of all continuous and categorical characteristics of the woman collected at baseline, the characteristics of the baby, birth complications, and characteristics of the physician will be performed. Any characteristics that are found to be statistically associated with (p\<0.05) the dependent variable (Screen positive) will be considered for inclusion in a multivariate logistic model.
3. Characteristics associated with screening for Type 2 DM or IGT: A univariate analysis of all continuous and categorical characteristics of the woman collected at baseline, the characteristics of the baby, birth complications, and characteristics of the physician will be performed. Any characteristics that are found to be statistically associated with (p\<0.05) the dependent variable (Screen positive) will be considered for inclusion in a multivariate logistic model.
4. Survey:

The family physicians and patients will be contacted for a semi-structured interview once the patient is at least six months postpartum. The investigator who is performing the interview will be blinded to the patient's group allocation. They will be surveyed to determine their satisfaction with the screening method and qualitatively review their suggestions to evaluate barriers to adoption of the screening recommendation and increase adoption in their practice or experience. (Appendix 1E) For qualitative responses, the responses will be tabulated and a percentage reported. For responses that are reported as a Likert scale, a mean and a standard deviation will be calculated. These results will be used to aid in the planning process of future studies.

Sample size:

The design is a 2x2 factorial design and we will power the study to detect a difference in proportions between the four groups. If we assume a power of 80% and an alpha error of 5% we would require 67 patients in each group to detect a difference of proportions of 17%, 20%, and 35% in the groups that received either intervention or both combined. We assumed that the combination of patient and physician letter would lead to higher screening rates. A larger sample size is required when an interaction is suspected to ensure that the study has sufficient power to detect an interaction.16 The total sample size of the study would be 268 patients.

From our previous study that evaluated the screening rate using a 2 hour 75g OGTT in the year 2000 at the Ottawa Hospital we found that 0% of patients were being screened. Conway et al14 in their population of women diagnosed with GDM were achieved a 17% screening rate when the the women were given the requisition for screening but no reminder letter outlining the importance of the screening test.

Feasibility Issues:

At the Ottawa Hospital 168 and 149 women who delivered and had GDM in 1997 and 2000 respectively. Averaging over the two years, we would estimate that approximately 160 women per year delivered at the Ottawa Hospital with GDM. Also, at the General campus of the Ottawa Hospital, approximately 30 patients per year are followed for GDM at the HRU but deliver at the Montfort hospital. Therefore, we would estimate that there would be a potential study population of 190 women at the Ottawa Hospital per year. We would also assume that 10% of women would decline to participate in the study, further reducing the potential population to 171 women. Therefore we would anticipate the study would take almost two years to complete enrollment.

Issues that may affect the success of this study are as follows: the acceptability of the study to the patient population of interest, whether the consent process itself would increase the screening rate as compared to the screening rate for the year 2000, and publication of new guidelines. It is unlikely that the consent process would be enough to increase screening rates alone as most studies on physician behaviour changes have shown that multiple interventions are required.11 The Clinical Practice Guideline Group of the CDA have started to meet to discuss an update of the guideline. Evidence published since the publication of the guidelines in 1998 lends more support to the recommendation to screen with a 2 hour OGTT.

Ethical consideration This study has received approval by the Ottawa Hospital Research Ethic Board. All records paper and electronic will be stored within the Hospital under the jurisdiction of the Cinical Epidemiogy Program of the Ottawa Health Research Institute. All records are confidential and only the Investigators and the Research Associate will have access to the records.

Knowledge Translation and dissemination plans The results of the study will be published in a peer-reviewed journal. The strategy that led to the highest proportion of patients being screened will be considered for adoption as part of routine clinical care at the Ottawa Hospital. The results of the survey will be utilized to improve the strategy if significant barriers to adoption were identified. The physician members of the team (Dr.'s Clark, Keely, and Karovitch) will be utilized as opinion leaders to encourage dissemination of the "best" screening strategy to other hospitals and physician groups involved in the care of patients with gestational diabetes.