Dietz, Patricia M. DrPH, MPH1; Vesco, Kimberly K. MD, MPH2; Callaghan, William M. MD MPH1; Bachman, Donald J. MS2; Bruce, F Carol MPH1; Berg, Cynthia J. MD, MPH1; England, Lucinda J. MD, MPH1; Hornbrook, Mark C. PhD2
Gestational diabetes mellitus (GDM), a carbohydrate intolerance that is first recognized during pregnancy, strongly predicts future development of type 2 diabetes. Immediately after delivery, approximately 2% to 14% of women with GDM will continue to have glucose intolerance in the diabetic range, and another 3–35% will have impaired fasting glucose or glucose tolerance.1–4 The American College of Obstetrics and Gynecologists and American Diabetes Association recommend postpartum diabetes screening for women with GDM-affected pregnancies.5,6 Women with GDM require a postpartum glucose test to identify ongoing abnormal glycemia so that lifestyle and/or pharmacologic interventions can be implemented. Without intervention, up to 50% of women with GDM will develop type 2 diabetes within 5 years after delivery.7
Previous studies of postpartum fasting glucose or glucose tolerance testing indicate that testing rates are suboptimal, ranging from 188 to 57%.4 However, recent increases in GDM prevalence and growing awareness among clinicians of its associated risk of type 2 diabetes,9 may have resulted in improvements in postpartum test rates. We examined postpartum screening rates from 1999 to 2006 in a health maintenance organization (HMO) that implemented a protocol for universal postpartum screening among women with GDM during the study period. We also assessed whether practice site or patient characteristics were associated with clinician orders for a postpartum glucose test and with completion of those orders.
MATERIALS AND METHODS
We analyzed data from Kaiser Permanente Northwest, a large nonprofit prepaid, federally-certified, Joint Commission-accredited, group practice HMO with 470,000 members in western Oregon and Washington State. We selected all pregnancies ending in live birth or stillbirth between January 1, 1999, and December 31, 2006, for which the mother was still enrolled in the HMO 3 months after delivery. Most members received prenatal care at one of eight practice sites, one of which serves as a maternal–fetal medicine referral practice. At each provider site, the annual number of patients receiving prenatal care ranged from 459 to 867.
To identify pregnancies, we used an ongoing validated computer algorithm that accessed multiple Kaiser Permanente Northwest automated data systems. Details of this algorithm and examples of its applicability have been previously published.10 Briefly, the algorithm searches hospital discharge abstracts; prenatal outcome data, and diagnostic and procedure codes from same-day surgery records, ambulatory encounters, and emergency department and home health visits; outside professional and facility claims and referrals; imaging procedures; laboratory test results; and pharmacy dispensing records. The resulting database identifies pregnancy “episodes” and includes all health care encounters with their attendant ancillary diagnoses, procedures, and other measures of health services. After pregnancies are identified, a probabilistic method is used to match birth certificates to the Kaiser Permanente Northwest data to gather additional demographic information. This study was approved by the Centers for Disease Control and Prevention and the Kaiser Permanente Northwest Institutional Review Boards.
During the study period, screening for GDM was a component of usual care for pregnant women at Kaiser Permanente Northwest. Unless clinicians suspected GDM early in the pregnancy, women were generally screened between 24 and 28 weeks gestation with a 50-g, 1-hour glucose challenge test. Those with 1-hour values 140 mg/dL or more were offered a 100-g, 3-hour oral glucose tolerance test (OGTT). We considered “confirmed” GDM cases to be those with positive laboratory test results, or an International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) code of 648.8 (“abnormal glucose tolerance”) combined with a dispense for insulin or glyburide. Positive laboratory test results included fasting blood sugar of 126 mg/dL or more (alone or as part of a 3-hour 100-g OGTT) or 1-hour glucose challenge test of 200 mg/dL or more or at least two abnormal values in the 3-hour 100-g OGTT (fasting more than 105 mg/dL, 1-hour more than 190 mg/dL, 2-hour more than 165 mg/dL, 3 hour more than 145 mg/dL). To validate these methods, a subset of medical records (n=130) was reviewed by two coauthors (K.V. and D.B.). During the medical chart review, additional criteria were identified for a group of “unconfirmed” cases. We refer to these cases as “unconfirmed” GDM because medical chart review found many were cases but some were not. These cases were characterized by having an ICD-9-CM code of 648.8 during the pregnancy and at least one abnormal value in the 3-hour 100-g OGTT and no dispense for insulin or glyburide. We could not find a systematic way to separate the “true” GDM cases and therefore report results separately for confirmed and unconfirmed cases of GDM.
We assessed two outcome variables: a clinician order for either a fasting plasma glucose test (FPG) or 2-hour 75-g OGTT within 1 month before to 3 months after delivery and test results in the woman’s medical record indicating that she completed the test within 3 months of delivery. The standard protocol at Kaiser Permanente Northwest is to order a FPG postpartum. To address potential provider variation, we searched for both tests and found that no postpartum 2-hour 75-g OGTT test was ordered during the study period. Test results were examined and the percentage of women with diabetes (fasting blood glucose of 126 mg/dL or more) or with impaired fasting glucose (fasting sugar 100–125 mg/dL) were calculated.
We assessed associations between the outcome variables and patient characteristics and practice site. Maternal educational attainment, parity, marital status, race/ethnicity, and trimester of entry into prenatal care were obtained from birth certificate records. Maternal age, prepregnancy body mass index, use of insulin or glyburide during pregnancy, the woman’s primary language, clinical visit with a nutritionist, infant birth weight, postpartum visit within 3 months of delivery, and practice site where the woman received prenatal care were obtained from Kaiser Permanente Northwest records.
Approximately 20% of the women who delivered a live birth or stillbirth during 1999–2006 had more than one delivery during the study period, as did 25% of women with a GDM-affected pregnancy. Therefore, for the trend analyses, generalized estimating equations were used to account for the correlation between pregnancies in the same woman. Logistic regression was used to test for the statistical significance in three separate trends from 1999 through 2006: 1) the prevalence of pregnancies with GDM; 2) the prevalence of pregnancies with laboratory orders for a postpartum glucose test; and 3) the prevalence of completed postpartum tests. For each model, year was entered as an ordinal variable. When assessing whether practice site or patient factors were associated with clinician orders and completed tests, we restricted the analysis to 2004–2006 to investigate the most recent behaviors and practices. We ran χ2 tests for independence to test for differences using generalized estimating equations to account for the correlation between pregnancies in the same woman. Because glucose tests cannot be performed without an order, we restricted our analysis of factors associated with completion of the postpartum test to women who had an order. Logistic regression with a backwards elimination method was used to assess the independent associations of practice site and maternal characteristics with clinician orders for a postpartum test and for completed tests. Missing observations on the patient characteristics were deleted from the models (missing n=35 of 461, 7.6% for clinician orders; and n=13 of 356, 3.7% for completed tests). Statistical analyses were run in Statistical Analysis Software (SAS, SAS Institute Inc., Cary, NC).
Among 36,251 eligible pregnancies ending in live births or stillbirths during 1999 through 2006, 1,127 (3.1%) confirmed cases and 205 (0.6%) unconfirmed cases of GDM were identified. Most women with confirmed GDM had greater than 12 years of education, were between 20 and 34 years of age, and were married. Approximately 60% were white and 21% were Asian. More than 38% were nulliparous.
During 1999–2006, the percentage of pregnant women identified with confirmed GDM increased from 2.9% in 1999 to 3.6% in 2006 (P<.01) (Fig. 1). Unconfirmed cases increased from 0.4% to 0.7% (data not shown). Laboratory orders for postpartum FPGs increased during the study period, from 15.9% in 1999 to 79.3% in 2004 (P<.01), and then remained stable through 2006 at 74.0% in 2006 among confirmed cases (Fig. 2). Laboratory orders for unconfirmed cases increased from 5.0% in 1999 to 62.9% in 2006 (data not shown). The percentage of women with confirmed GDM who completed a FPG within 3 months of delivery increased from 9.0% in 1999 to 57.8% in 2004 (P<.01), and then remained stable through 2006 at 50.3% (Fig. 2). Completed glucose tests for unconfirmed cases increased from 5.0% in 1999 to 48.6% in 2006. Among women with confirmed GDM, 10.7% (40 of 375) had impaired fasting glucose, and 2.1% (8 of 375) were diagnosed with diabetes. Of those with unconfirmed GDM, 6.7% (3 of 45) had impaired fasting glucose, and none met the diagnosis for diabetes. Because none of the women received an OGTT, it is unknown what percent of women with GDM-affected pregnancies had impaired glucose tolerance postpartum.
Of all factors assessed for associations with clinician order for a postpartum glucose test during 2004–2006, the greatest variation was found for practice site. Among the eight practice sites, laboratory order rates for a postpartum glucose test ranged from 28.6% to 93.6% among confirmed cases. In addition, maternal characteristics associated with laboratory orders included attending the postpartum 6-week visit, education of 13 years or more, first birth, attending a nutrition visit during pregnancy, and entering prenatal care in the first trimester (Table 1). However, when all maternal characteristics and practice site were entered into a logistic regression model, practice site, education 13 years or greater, and entering prenatal care in the first trimester remained statistically significant (Table 2).
Among women with confirmed GDM and who received an order for a postpartum glucose test, completed postpartum tests ranged from 42.3% to 84.5% among the eight practice sites. Women with a completed test were more likely to attend the postpartum 6-week visit, be Asian or Hispanic, and to not speak English as their primary language (Table 3). When all maternal characteristics and practice site were entered into a logistic regression model, attending the postpartum visit (adjusted odds ratio [AOR] 8.00, 95% confidence interval [CI] 1.98–32.40) and being Asian (AOR 3.22, 95 % CI 1.67–6.18) or Hispanic (AOR 2.62, 95% CI 1.25–5.48) compared with being white remained statistically significant; practice site did not.
We found that, among women with a GDM-affected pregnancy, both postpartum glucose test orders as well as prevalence of completed glucose tests increased between 1999 and 2004 and then remained stable through 2006. However, even when rates were at their highest in 2004, 42% of women with confirmed GDM did not get a postpartum FPG, and orders were not placed for over one fifth of the women. The low postpartum test rate suggests that some women with diabetes or impaired fasting glucose remained unidentified. Of those tested, 10.7% had impaired fasting glucose, and 2.1% were diagnosed with diabetes. If these rates are applied to those not tested, we estimate that 80 women with impaired fasting glucose and 16 women with diabetes were missed. Because no postpartum OGTTs were ordered or completed, these estimates are conservative, because the OGTT is more sensitive than the FPG. For example, one study of 527 women with GDM found postpartum OGTTs identified 25 women with type 2 diabetes whereas the FPG identified 4, and the OGTT identified an additional 87 women with impaired glucose tolerance and normal fasting levels.11 Early diagnosis of diabetes mellitus is essential for preventing microvascular and macrovascular complications caused by long-standing, persistent hyperglycemia. In addition, women with impaired fasting glucose or impaired glucose tolerance can reduce their risk of developing type 2 diabetes by losing weight and exercising.12 Finally, for women with normal postpartum glucose test results, the postpartum visit provides an opportunity to educate them about lifestyle change and long-term follow up, because these women remain at increased risk for both type 2 diabetes and abnormal glucose homeostasis in subsequent pregnancies.
We identified patient and provider behaviors associated with postpartum glucose testing that are consistent with previous literature. As expected, patient attendance at the 6-week postpartum visit2,13 and providers’ submission of laboratory orders13 increased the likelihood that women would get a postpartum glucose test. In previous research, clinical factors, such as higher mean blood glucose values or insulin use during pregnancy, were also associated with increased testing in some,14 but not all studies.1,3,8 We did not find insulin or glyburide use to be associated with getting a postpartum test, perhaps because women taking these drugs were considered at higher risk, and were followed using other means, such as home blood sugar monitoring after delivery. We found that Asian or Hispanic women were more likely to get the test than white women, which is consistent with at least one other study.2 However, a clinician order was required for completing the postpartum test, and in our HMO study population, we found large variation in clinician orders by practice site.
The increase in submission of laboratory orders and in completed postpartum tests during the study period reflects changes in practice at Kaiser Permanente Northwest. In 2004, Kaiser Permanente Northwest implemented a new protocol, Management of Pregnant Women with Diabetes, to establish the standard for coordinated and consistent diabetes case management of pregnant women with GDM, preexisting diabetes mellitus and impaired glucose tolerance. As part of this protocol, women with GDM are referred to a Diabetes Team Nurse who is responsible for the coordination of her care during pregnancy through the 6-week postpartum visit. The nurse helps ensure that GDM patients have a standing order for a postpartum FPG and encourages them to complete the test before the 6-week visit. The increases in clinician orders began as the protocol was being developed and they peaked in 2004 when the protocol was implemented, and then remained stable. There were substantial differences in clinician orders by practice site. Based on informal discussion with nurses at each practice site, it seems that practice sites with systematic procedures for ordering the tests and with protected case-management time for nurses had higher clinician order rates. The findings of this study, overall, are limited to integrated staff model health maintenance organizations. It is unknown if similar rates would have been found in other health care settings.
The Kaiser Permanente clinical protocol for postpartum diabetes screening among women with GDM established that the FPG would be the standard for identifying persistent abnormal glycemia postpartum. Although the OGTT will detect a greater percent of women with glucose abnormalities, it is unknown whether women are less likely to obtain it compared with the FPG due to its inconvenience. Further investigation is warranted regarding which clinician order (FPG or OGTT) will result in the highest number of women identified with glucose abnormalities.
In our study, identification of GDM through ICD-9-CM codes proved to be a challenge. Through an iterative process of comparing GDM codes with medical record data, we discovered that the ICD-9-CM code 648.8 alone was not specific. It included women who failed the 1-hour glucose challenge test screen but had normal values on the 3-hour 100-g oral and no evidence of GDM. We then restricted our definition of GDM to at least two abnormal values in the 3-hour 100-g OGTT. However, through further medical record review, we learned this approach was insufficient because we found cases who were treated for GDM and who had had no 3-hour 100-g oral test. In these instances, the medical record showed that the clinician had reason to diagnose GDM without the test (eg, documented home monitoring). Thus, we added to the case definition those with an ICD-9-CM code 648.8 combined with a pharmacy dispense for insulin or glyburide. Medical chart review also indicated that an additional group of women with an ICD-9-CM code of 648.8 during the pregnancy and at least one abnormal value in the 3-hour 100-g OGTT included many GDM cases. However, we were unable to eliminate the noncases systematically and thus we made the conservative decision to report the results separately for this group. However, similar to the confirmed group, trends in orders and completed tests increased in the unconfirmed group. The difficulty in identifying GDM cases through ICD-9 CM codes, laboratory results, and dispenses for insulin and glyburide is not surprising given that the clinical definition of GDM, glucose intolerance first manifest during pregnancy, does not specify the means for diagnosis. It is unknown if similar rates would have been found in other health care settings.
Both the American College of Obstetrics and Gynecologists and American Diabetes Association have recommendations for postpartum diabetes screening for women with GDM-affected pregnancies.5,6 However, screening rates have been low, perhaps because providers are not familiar with recommendations, lack understanding about the implications of GDM beyond pregnancy, or perceive confusion among the recommendations for screening. Provider protocols that promote universal screening among women with GDM-affected pregnancies could address confusion among individual practitioners. This study found significant increases in clinician laboratory orders for postpartum glucose tests among women with prior GDM and subsequent completion of postpartum tests concurrent with the implementation of a clinical protocol for universal postpartum testing among women with GDM. The results suggest that significant improvements in postpartum testing can be achieved when clinical systems are designed to meet this goal.
1. Kim C, Tabaei BP, Burke R, McEwen LN, Lash RW, Johnson SL, et al. Missed opportunities for type 2 diabetes mellitus screening among women with a history of gestational diabetes mellitus. Am J Public Health 2006;96:1643–8.
2. Russell MA, Phipps MG, Olson CL, Welch HG, Carpenter MW. Rates of postpartum glucose testing after gestational diabetes mellitus. Obstet Gynecol 2006;108:1456–62.
3. Schaefer-Graf UM, Buchanan TA, Xiang AH, Peters RK, Kjos SL. Clinical predictors for a high risk for the development of diabetes mellitus in the early puerperium in women with recent gestational diabetes mellitus. Am J Obstet Gynecol 2002;186:751–6.
4. Hunt KJ, Conway DL. Who returns for postpartum glucose screening following gestational diabetes mellitus? Am J Obstet Gynecol 2008;198:404.e1–6.
5. American College of Obstetricians and Gynecologists Committee on Practice Bulletins–Obstetrics. ACOG Practice Bulletin. Clinical management guidelines for obstetrician-gynecologists. Number 30, September 2001 (replaces Technical Bulletin Number 200, December 1994). Gestational diabetes. Obstet Gynecol 2001;98:525–38.
6. American Diabetes Association. Gestational diabetes mellitus. Diabetes Care 2004;27:S88–90.
7. Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type 2 diabetes: a systematic review. Diabetes Care 2002;25:1862–8.
8. Conway DL, Langer O. Effects of new criteria for type 2 diabetes on the rate of postpartum glucose intolerance in women with gestational diabetes. Am J Obstet Gynecol 1999;181:610–4.
9. Ferrara A. Increasing prevalence of gestational diabetes mellitus: a public health perspective [published erratum appears in Diabetes Care 2007;30:3154]. Diabetes Care 2007;30:S141–6.
10. Hornbrook MC, Whitlock EP, Berg CJ, Callaghan WM, Bachman DJ, Gold R, et al. Development of an algorithm to identify pregnancy episodes in an integrated health care delivery system. Health Serv Res 2007;42:908–27.
11. Kitzmiller JL, Dang-Kilduff L, Taslimi MM. Gestational diabetes after delivery. Short-term management and long-term risks [published erratum appears in Diabetes Care 2007;30:3154]. Diabetes Care 30:S225–35.
12. Knowler WC, Barrett-Connor E, Fowler SE, Hamman RF, Lachin JM, Walker EA, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 2002;346:393–403.
13. Greenberg LR. Gestational diabetes mellitus: antenatal variables as predictors of postpartum glucose intolerance. Obstet Gynecol 1995;86:97–101.
14. Smirnakis KV, Chasan-Taber L, Wolf M, Markenson G, Ecker JL, Thadhani R. Postpartum diabetes screening in women with a history of gestational diabetes. Obstet Gynecol 2005;106:1297–303.