Several clinical and immunologic factors have been associated with an increased risk for postpartum thyroid dysfunction. These factors include a history of medically treated Graves disease in remission before the pregnancy [4], a previous history of postpartum thyroiditis [5], a family history of thyroid disease [6], the presence of the HLA marker DR4 [7, 8], and the presence of antimicrosomal antibodies at term or before delivery [9-11].

Approximately 20% of patients with type I diabetes have immunologic evidence of autoimmune thyroid disease [12], and a substantial proportion of these patients have or develop thyroid dysfunction during their lifetime [13]. A previous study, which had complete follow-up in only 57% of the original cohort of consecutive type I diabetic women giving birth, showed an observed annual incidence of postpartum thyroiditis of 10.5% [14]. The low follow-up rate raised questions about the general applicability of the study's findings to diabetic women in the postpartum period. Moreover, the clinical effect of postpartum thyroid dysfunction with respect to symptoms and diabetes control was not assessed.

We did a prospective cohort study to determine the incidence and the clinical effect of, as well as risk factors for, postpartum thyroid dysfunction in patients with type I diabetes who delivered a child during a 2-year period in a geographically defined area.

Methods

Top Methods Results Discussion Author & Article Info References

The study was done at McMaster University Medical Centre, which is the only high-risk pregnancy center in the Hamilton, Ontario catchment area, serving a population of approximately 500 000 people. More than 90% of all pregnant women with type I diabetes who live in the area deliver at this hospital, and most are also followed prenatally in the Medical Centre's ambulatory diabetes clinic.

Between May 1989 and June 1991, all 51 women with insulin-dependent diabetes mellitus diagnosed before the age of 36 years and at least 18 months before delivery were asked to participate in this study within 1 week of having their babies. Our patient group represented approximately 0.33% of the 15 435 women giving birth in Hamilton hospitals during this 2-year period. None of these patients was taking thyroid hormone or thyroactive or immunosuppressive drugs at the time of delivery. Patients who gave informed consent were followed prospectively for 6 months; all clinical and biochemical assessments were done at baseline (within the first postpartum week) and then again 3 and 6 months after delivery.

Clinical assessments included an examination of the thyroid by one of three experienced endocrinologists. In addition, all patients were asked to complete a thyroid symptom checklist (before having their blood drawn for thyroid testing) by answering "yes" or "no" to a set of 15 questions dealing with symptoms of hypothyroidism and thyrotoxicosis. These questions, derived from previous lists of thyroid symptoms [15, 16], asked about low energy, increased fatigue, difficulty with concentration or memory, sleep disturbance, cold or heat intolerance, dry or sweaty skin, shakiness, palpitations, shortness of breath, altered bowel habit, difficulty in climbing stairs, and muscle cramps. Diabetes control was assessed at each visit by obtaining the appropriate history and measuring glycated hemoglobin and fructosamine.

Thyroid tests done in the clinical laboratory included total thyroxine (T4) by radioimmunoassay (International Diagnostic Services, Scarborough, Ontario; normal range, 60 to 145 nmol/L); thyroxine-binding globulin (TBG) by radioimmunometric assay (Ciba Corning Canada Inc., Richmond Hill, Ontario; normal range, 150 to 360 nmol/L); and thyroid-stimulating hormone (TSH) by a sensitive two-site fluoroimmunometric assay (Pharmacia Canada, Montreal, Quebec) with a lower detection limit of 0.07 mU/L and an interassay precision of 7.9% in the normal range (normal range, 0.5 to 5.5 mU/L). Thyroid microsomal antibodies were measured by agglutination (Miles Canada Inc., Etobicoke, Ontario); a titer of 1:100 or more was considered positive. Glycated hemoglobin was measured by affinity-column chromatography (Iso Lab Inc., Akron, Ohio; normal range, 4.5% to 7.5%), and fructosamine was measured using two different kits (Hoffman-La Roche Ltd., Etobicoke, Ontario, before June 1990 and Boehringer Mannheim Diagnostics, Montreal, Quebec after 1 June, 1990; normal range, 205 to 285 µmol/L).

Diagnostic criteria for thyroid dysfunction were defined before the study and included overt hypothyroidism (TSH > 5.5 mU/L, T4 < 60 nmol/L, and a low T4/TBG ratio); subclinical hypothyroidism (TSH > 5.5 mU/L, T4 of 60 to 145 nmol/L, and a normal T4/TBG ratio), and thyrotoxicosis (TSH < 0.1 mU/L, T4 > 145 nmol/L and a high T4/TBG ratio). Tests were done at the 3-month and 6-month postpartum visits. Patients who developed overt hypothyroidism were treated with thyroid hormone and classified as having postpartum thyroiditis. Patients with persistent thyrotoxicosis underwent technetium scanning and measurement of radioiodine uptake; Graves disease was diagnosed when uptake was elevated and the scan was diffuse, and postpartum thyroiditis was diagnosed when the uptake was suppressed.

Statistical Analysis

Exact 95% CIs were calculated for all estimates of incidence. Characteristics of eligible patients, changes in symptom scores, and indices of diabetes control were compared using the Mann-Whitney U statistic. Risk factors for postpartum thyroid dysfunction were compared using chi-square or Fisher exact tests.

Results

Top Methods Results Discussion Author & Article Info References

Fifty-one patients were eligible for evaluation during the study period. We excluded those who refused to consent (4 patients), did not attend both of the postpartum visits (6 patients), or had difficult venous access (1 patient). Baseline characteristics of the remaining 40 participants and of the 11 nonparticipants are presented in Table 1. The two groups differed with respect to gravidity (P = 0.005), weeks of gestation at the time of delivery (P = 0.03), and glycated hemoglobin level (P = 0.05).

Postpartum thyroid dysfunction occurred in 10 of 40 patients (25%; 95% CI, 12.7% to 41.2%) within the first 6 months after delivery. Postpartum Graves disease developed in 1 patient and postpartum thyroiditis developed in 9 patients. The types of thyroid dysfunction and the time of occurrence in these 10 patients are summarized in Table 2. Severity ranged from subclinical hypothyroidism (TSH, 9.2 mU/L; T₄, 128 nmol/L, T_4/TBG, 0.40) to profound hypothyroidism (TSH, 216 mU/L; T₄ < 10 nmol/L; T_4/TBG, < 0.03). Thyrotoxic patients had T₄ values ranging from 145 to 267 nmol/L and T_4/TBG ratios ranging from 0.7 to 0.96; TSH was undetectable in all thyrotoxic patients.

Baseline thyroid test results were available in all patients except two; one of these was euthyroid at both postpartum visits, and one subsequently developed postpartum thyroiditis. This latter patient had documented normal thyroid test results 10 weeks before delivery. Subclinical hypothyroidism was detected at baseline in five patients, four of whom subsequently developed postpartum thyroiditis. Two of these four patients were thyrotoxic 3 months after delivery and hypothyroid 6 months after delivery. The other two patients had normal thyroid test results at 3 months; however, by 6 months after delivery, one patient had overt hypothyroidism and the other had thyrotoxicosis. Of the 34 patients who were euthyroid during the first postpartum week (that is, at baseline), 5 (14.7%; 95% CI, 5.0% to 31.1%) developed postpartum thyroiditis and 1 (2.9%; 95% CI, 0.1% to 15.3%) developed postpartum Graves disease during the first 6 months after delivery.

No differences in baseline thyroid tests, glycated hemoglobin, fructosamine, gravidity, birth weight, maternal age, and weeks of gestation were found between patients who developed thyroid dysfunction and those who did not (data not shown). The frequency of other baseline characteristics in patients who did and did not develop thyroid dysfunction is shown in Table 3. No patient with a normal thyroid gland (by palpation) at baseline developed postpartum thyroid dysfunction during 6 months of follow-up. At baseline, 20 patients had palpably abnormal thyroid glands (goiters) that were judged to be moderately or very enlarged, firmer than normal in consistency, or both; none of these was nodular. Of these 20 patients, 9 developed postpartum thyroid dysfunction (all 4 patients with subclinical hypothyroidism at baseline and 5 of 16 patients without subclinical hypothyroidism [P = 0.03]; data not shown). Among patients without baseline subclinical hypothyroidism, 4 of 5 patients had detectable antimicrosomal antibodies at baseline and 1 of 11 did not develop postpartum thyroid dysfunction (P = 0.01; data not shown).

To determine the extent to which thyroid dysfunction was associated with symptoms of thyroid disease, the number of positive responses to the symptom checklist was counted at each visit. Because many of these symptoms are nonspecific and could be related to the recent pregnancy, to diabetes, or to the general stress of the postpartum period, the number of positive responses obtained during the first postpartum week was subtracted from the number obtained at the 3- and 6-month visits for each patient. The changes from baseline in total symptom count, hypothyroid symptom count, and thyrotoxic symptom count for euthyroid patients and for those who developed thyroid dysfunction were compared at each postpartum time interval (Figure 1). Patients with postpartum thyroid dysfunction (either thyrotoxic or hypothyroid) had a greater change in symptom score than did euthyroid patients at both the first (P = 0.04) and second (P = 0.009) postpartum visits. Patients with hypothyroidism did not differ from euthyroid patients regarding change in symptom score at the first postpartum visit (P > 0.2); however, at the second postpartum visit, patients with hypothyroidism did have a greater change in symptom score (P = 0.02) than euthyroid patients (see Figure 1).

View larger version (19K): [in this window] [in a new window]   Figure 1. Change in symptom score from baseline in patients with and without thyroid dysfunction. The change in symptom score for euthyroid patients at each postpartum visit is indicated by circles (bars represent 95% CIs). The change in symptom score for each patient who had thyroid dysfunction at either the 3- or 6-month visit is indicated by an "X." P values were determined using the Mann-Whitney U test. Patients who developed hypothyroidism by the 6-month postpartum visit had a greater change in hypothyroid symptom score than did the euthyroid patients at this time. Patients who had thyroid dysfunction (either hypothyroidism or thyrotoxicosis) had a greater change in total symptom score than did the euthyroid patients at both the 3- and 6- month visits.

Postpartum thyroid dysfunction had no statistically significant effect on indices of glycemic control, including glycated hemoglobin concentration, serum fructosamine level, or the number of hypoglycemic reactions at either postpartum interval (P > 0.2 for all comparisons).

The potential effect of the nonparticipating patients on the estimate of the incidence of postpartum thyroid dysfunction was assessed in two ways. First, because participants and nonparticipants differed at baseline with respect to glycated hemoglobin level, gravidity, and weeks of gestation, we tested (chi-square test) for associations between these variables and thyroid dysfunction in the participants; no associations were found (P = 0.17, P > 0.2 and P > 0.2, respectively); this absence of associations suggests that the differences exhibited by the nonparticipants would have a negligible effect on their incidence of thyroid dysfunction. Second, when a sensitivity analysis was done in which all nonparticipants were assumed not to have developed thyroid dysfunction (that is, 10 of 51 patients), the incidence of postpartum thyroid dysfunction was 19.6% (95% CI, 9.8% to 33.1%). Therefore, even if nonparticipants did have a lower risk for subsequent thyroid dysfunction than participants, the incidence in type I diabetic patients would still be substantial. Conversely, assuming that all of the patients in this group developed postpartum thyroid dysfunction (21 of 51 patients), the incidence would be even higher (41.2%; 95% CI, 27.6% to 55.8%).

Discussion

Top Methods Results Discussion Author & Article Info References

In our study, 25% of patients with type I diabetes developed thyroid dysfunction in the first 6 months after delivery. Postpartum thyroiditis (incidence, 22.5%) was the most common cause of thyroid dysfunction and occurred as either transient thyrotoxicosis, hypothyroidism, or transient thyrotoxicosis followed by hypothyroidism.

Subclinical hypothyroidism was present at baseline in four patients who subsequently developed postpartum thyroiditis. This finding, in addition to the strong association of goiter and the presence of antimicrosomal antibodies with subsequent thyroid dysfunction, suggests that autoimmune thyroid damage may be present even at the time of delivery and may be intensified in the postpartum period. Moreover, a high TSH may amplify or accelerate the autoimmune process [17, 18]. This observation also illustrates a further clinical pattern of postpartum thyroiditis: subclinical hypothyroidism, followed by either thyrotoxicosis or hypothyroidism. Although such a pattern has been described previously [10], it was common (4 of 9 patients) in our patients with type I diabetes.

The high incidence of postpartum thyroid dysfunction in patients with type I diabetes is consistent with the well-recognized association between nonpostpartum autoimmune thyroid disease and type I diabetes [12, 13] and suggests that type I diabetes is an important risk factor for postpartum thyroid disease. This finding and the association of thyroid dysfunction with goiter and the presence of thyroid microsomal antibodies at delivery support the view that pregnancy and the postpartum period trigger the development or acceleration of autoimmune thyroid disease in predisposed persons. That many patients with a history of postpartum thyroid disease eventually develop permanent hypothyroidism further supports this concept [19-21].

The clinical significance of this association is reflected in the observation that affected patients experienced more symptoms of thyroid dysfunction than did unaffected patients, even though affected patients were identified early in the course of thyroid disease through screening, were experiencing the stresses associated with a new baby, and were continuing to deal with diabetes. This finding is consistent with other studies suggesting that postpartum thyroid disease produces symptoms that may affect quality of life [9, 22, 23]; thus a strategy of careful follow-up and early treatment of affected patients is recommended. Thyroid dysfunction did not appear to adversely affect diabetes control, but this may have been due to early detection and treatment.

Patients with type I diabetes may therefore benefit from routine screening for thyroid dysfunction at postpartum visits and from regular follow-up of any abnormal results. Moreover, further studies of these patients may yield clinical, immunologic, and biochemical insights into the pathogenesis of autoimmune thyroid disease.

Abbreviation

TBG: thyroid-binding globulin

TSH: thyroid-stimulating hormone

T₄ : thyroxine