diabetes research and clinical practice 1 4 3 (2 0 1 8) 1 5 1–15 8

Contents available at ScienceDirect

Diabetes Research
and Clinical Practice
journal homepage: www.elsevier.com/locat e/dia bre s

Type 1 diabetes impairs female fertility even before

it is diagnosed

Yung-Hsiang Lin a, Ko-Jung Chen b, Yun-Shing Peng a, Pau-Chung Chen c,d,

Yao-Hsu Yang b,e,*
a
Division of Endocrinology and Metabolism, Chang Gung Memorial Hospital Chiayi Branch, Chiayi County, Taiwan
b
Health Information and Epidemiology Laboratory of Chang Gung, Memorial Hospital Chiayi Branch, Chiayi County, Taiwan
c
Department of Environmental and Occupational Medicine, National Taiwan University Hospital and National Taiwan University College of
Medicine, Taipei, Taiwan
d
Institute of Occupational Medicine and Industrial Hygiene, National Taiwan University College of Public Health, Taipei, Taiwan
e
Department of Traditional Chinese Medicine, Chang Gung Memorial, Hospital Chiayi Branch, Chiayi County, Taiwan and School
of Traditional Chinese Medicine, College of Medicine, Chang Gung University, Taoyuan, Taiwan

A R T I C L E I N F O A B S T R A C T

Article history: Aims: The aim of this study was to assess the fertility rate in Taiwanese women before and
Received 18 October 2017 after a diagnosis of type 1 diabetes. The potential risk factors which may have influenced
Received in revised form fertility were also investigated.
21 February 2018 Methods: We conducted this retrospective, nationwide, population-based, matched cohort
Accepted 3 July 2018 study using data from the Taiwan National Health Insurance Research Database. The dia-
Available online 9 July 2018 betic group (n = 1191) included women with type 1 diabetes aged between 16 and 30 years
in 2000. The non-diabetic group (n = 4764) was matched by sex, gender, income, and urban-
ization. The endpoints, including live births, abortions, and fertility, were tracked until the
Keywords:
end of 2013. Poisson regression was used to assess incidence rate ratios (IRRs). We also ana-
Type 1 diabetes
lyzed the influence of autoimmune thyroid disease, diabetic ketoacidosis, diabetic compli-
Fertility
cations, and daily insulin dose on the endpoints.
Hyperthyroidism
Results: The diabetic group had a lower rate of live births (IRR 0.67 [95% CI 0.62–0.73]) than
Hypothyroidism
the non-diabetic group, and it was even lower when combined with hyperthyroidism (IRR
Diabetic ketoacidosis
0.54 [0.39–0.74]). There were also fewer live births before a diagnosis of diabetes than after a
Polycystic ovarian syndrome
diagnosis of diabetes (IRR 0.58 [0.52–0.65] vs. 0.80 [0.71–0.90]). Diabetic ketoacidosis and a
higher daily insulin dose were strongly associated with abortion. Diabetic complications
significantly reduced the number of live births.
Conclusions: Type 1 diabetes compromises female fertility, even before it is diagnosed. Asso-
ciated hyperthyroidism further reduces fertility. Blood glucose and thyroid function
surveillance in infertile females may allow for an early diagnosis of type 1 diabetes and
associated thyroid disease.
Ó 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

* Corresponding author at: Chang Gung Memorial Hospital Chiayi Branch, No. 6, W. Sec., Jiapu Rd., Puzih City, Chiayi County 613,
Taiwan.
E-mail addresses: pchen@ntu.edu.tw (P.-C. Chen), r95841012@ntu.edu.tw (Y.-H. Yang).
https://doi.org/10.1016/j.diabres.2018.07.010
0168-8227/Ó 2018 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
152 diabetes research and clinical practice
1. Introduction
Type 1 diabetes mellitus (DM) is an autoimmune disease that
destroys b cells, and which requires lifelong insulin replace-
ment therapy. Hyperglycemia, insulin deficiency, and iatro-
genic hyperinsulinemia can induce hypogonadism and
hyperandrogenism, which can result in decreased fertility
[1]. Several cohort studies have also confirmed that women
with type 1 DM have fewer offspring and a higher rate of con-
genital abnormalities than healthy women [2,3] or their sib-
lings [4]. However, no study has investigated the fertility of
Asian women with type 1 DM, and the effect of autoimmunity
on fertility before type 1 DM is diagnosed remains unknown.
Type 1 DM is frequently associated with other autoim-
mune diseases such as autoimmune thyroid disease (ATD,
15–30%), celiac disease (4–9%), and Addison’s disease (0.5%)
[5]. Autoimmune thyroiditis can also decrease fertility and
induce pregnancy loss, even in euthyroid patients with anti-
bodies alone [6]. However, previous studies on fertility in
patients with type 1 DM have not addressed this issue.
In Taiwan, there were 7255 incident cases of type 1 DM from
1999 to 2010, of whom 52% were female. The incidence
increased from 1999 to 2010 with a higher age-specific inci-
dence rate in the younger groups (<30 years), and over half of
the female patients were of reproductive age (15–44 years) [7].
The aim of this study was to assess the fertility rate before
and after a diagnosis of type 1 DM. We also investigated the
effect of ATD and other risk factors which may have influenced
the fertility rate in this nationwide retrospective cohort study.
2. Material and methods
2.1. Source of datasets
The data in this study were obtained from the National Health
Insurance Research Database (NHIRD). The National Health
Insurance (NHI) program is a single payer program launched
in Taiwan on March 1, 1995, and it had a coverage rate of
around 99.9% of the total population by the end of 2014 [8].
The NHIRD contains all inpatient and ambulatory medical
claims data from the NHI program. Its completeness and
accuracy has been reported to be acceptable, and it has been
used in research studies of type 1 DM [9]. This study was
approved by the Institutional Review Board (IRB) of Chang
Gung Medical Foundation (IRB No. 201701132B0).
2.2. Matched Cohort Design
We searched for female patients with type 1 DM (Interna-
tional Classification of Diseases 9th Revision, Clinical Modifi-
cation [ICD-9-CM] codes 250.x1 or 250.x3) in the NHIRD from
2000 to 2013. Patients with type 1 DM who had catastrophic
illness/injury certificates (CICs) were defined as the case sub-
jects. In Taiwan, the National Health Insurance Administra-
tion (NHIA) issues a CIC to patients with type 1 DM.
Specialists confirm the diagnosis according to clinical mani-
festations, biochemistry data, and positive autoantibodies
before issuing a CIC. A previous study reported that the posi-
tive predictive value of a diagnosis of type 1 DM with a CIC
was 98.5% [7]. Case subjects younger than 16 years or older
1 4 3 ( 2 0 1 8 ) 1 5 1 –1 5 8
than 30 years in 2000 were excluded to ensure that the study
subjects were all within reproductive age during the study
period. The date when the CIC was issued was regarded as
the date of a diagnosis of type 1 DM in this study.
The matched patients were obtained from the Longitudinal
Health Insurance Database of 2005 (LHID-2005) matched in a
one-to-four ratio for age, gender, income, and urbanization
without type 1 DM. The LHID is a subset of the NHIRD which
contains one million randomly sampled beneficiaries from
2005. It accounts for about 5% of the total population in Taiwan
with no significant differences in gender distribution com-
pared to the NHIRD [8]. Income was classified as above or below
average (New Taiwan Dollars 15,840 per month), and urbaniza-
tion was categorized into city and villages, which may have
affected the accessibility and availability of medical care.
2.3. End-points and risk factors
The study end-points included live birth, abortion, and fertil-
ity. Live birth was defined as an inpatient diagnosis of normal
vaginal delivery with or without complications (NHI codes
81017C, 81024C, 81034C, 97001A, 97001K, 97002A, 97003B,
97004C, 97005D, 97931K, 97932A, 97933B, and 97934C), cesar-
ean delivery (NHI codes 81004C, 81005B, 81005C, 81028C,
81029C, 97006K, 97007A, 97008B, 97009C, 97014C, 98001K,
98002A, 98003B, and 98004C), and multifetal pregnancy (NHI
codes: 81018C, 81019C, 81025C, and 81026C) [10].
Abortion was defined as ICD-9-CM codes 632 (missed abor-
tion), 633 (ectopic pregnancy), 634 (spontaneous abortion),
and 637 (unspecified abortion) in the top five diagnoses of
inpatient and ambulatory claims [11]. The diagnosis of abor-
tion with an interval of more than 1 month, and 10 months
for an ectopic pregnancy, were regarded as separate episodes.
Fertility was defined as the sum of the number of live births
and abortions. The endpoints were tracked until the end of
the study period or death.
Risk factors that may have affected fertility were also stud-
ied, including ATD, diabetic ketoacidosis (DKA), diabetic com-
plications, and daily insulin dose. ATD, including
hyperthyroidism and hypothyroidism, was defined as the use
of thyroid or anti-thyroid preparations available in Taiwan.
Anatomical Therapeutic Chemical (ATC) codes were used to
identify drugs in the NHIRD (levothyroxine sodium H03AA01,
propylthiouracil H03BA02, carbimazole H03BB01, methima-
zole H03BB02). The ATC code is a drug classification system
developed by the World Health Organization (WHO) used for
research on drug utilization. Defined daily dose (DDD), which
was also developed by the WHO, is the assumed average main-
tenance dose per day for a drug used for its main indication in
adults [12]. We assumed that subjects with a DDD 28 of thy-
roid or anti-thyroid preparations had hypo- or hyperthy-
roidism. Patients who met both criteria probably had either
Hashimoto’s thyroiditis or post-surgical/radioiodine hypothy-
roidism, and were thus classified as having hypothyroidism.
We searched for DKA (ICD-9-CM code 250.1) in the top five
diagnoses of inpatient claims, and the sum of DKA episodes
after DM had been diagnosed was defined as diabetic control
status. Diabetic complications were evaluated using the
adapted Diabetes Complications Severity Index (aDCSI) [13].
The DCSI was first developed in 2008 and has been shown
 diabetes research and clinical practice 1 4 3 ( 2 0 1 8 ) 1 5 1 –1 5 8 153

to have a better ability to predict mortality than simply a

Table 1 – Characteristics of the study subjects.
count of complications [14]. The DCSI is a composite of select
laboratory results and diagnosed complications including car- Variables DM Non-DM p value
diovascular disease, nephropathy, retinopathy, peripheral (N = 1191) (N = 4764)
vascular disease, stroke, neuropathy, and metabolic. The n % n %
aDCSI was modified from the DCSI by excluding laboratory
Urbanization level 1.0000
data, and it has been validated in claims data [13]. It has been Villages 324 27.20 1296 27.20
shown to be a good measure of diabetes severity, and to have City 867 72.80 3468 72.80
a similar performance in predicting hospitalization compared Income (NTD) 1.0000
to the DCSI. A change in the aDCSI, which also reflects dia- 0–15840 1007 84.55 4028 84.55
betes control status, is defined as an increase in the score of >15841 184 15.45 736 15.45
Fertility <0.0001
less or more than 0.5 per year. Data on the dose of insulin
0 734 61.63 2281 47.88
(ATC code A10A) were also obtained from the NHIRD and cat- 1 258 21.66 1119 23.49
egorized into tertiles according to average daily dose. 2 164 13.77 1065 22.36
We found no cases of celiac disease (ICD-9-CM code 579.0) 3 35 2.94 299 6.28
in the study group, and no specific ICD-9-CM code could be
Comorbidities
used for Addison’s disease. Therefore these two diseases were Female infertility 0.8082
not analyzed. Female infertility (ICD-9-CM code 628) within Yes 75 6.30 291 6.11
the top five diagnoses in any inpatient or outpatient record No 1116 93.70 4473 93.89
was documented as a confounding factor. Hyperthyroidism <0.0001
Yes 148 12.43 78 1.64
2.4. Statistical analyses No 1043 87.57 4686 98.36
Hypothyroidism <0.0001
Yes 49 4.11 28 0.59
The distribution of demographic factors and the proportions of No 1142 95.89 4736 99.41
comorbidities between the patients with and without type 1
NTD; New Taiwan Dollars.
DM were compared. Poisson regression was used to assess inci-
dence rate ratios (IRRs) and the corresponding 95% confidence
intervals (95% CIs) of live births, abortions, and fertility for the
group, and this was even lower if combined with hyperthy-
time periods before and after a diagnosis of type 1 DM. We also
roidism (IRR 0.54, 95% CI 0.39–0.74, p = 0.0001). The abortion
calculated the IRRs (95% CIs) of live births, abortions, and fertil-
rate was similar between the DM and non-DM groups
ity stratified by groups according to hyperthyroidism and
whether or not they had ATDs. Similar to live births, fertility
hypothyroidism. To investigate the influence of DKA, aDCSI,
was lower in the DM group (IRR 0.76, 95% CI 0.71–0.81, p < 0.
and insulin dose on fertility in the female patients with type
0001), and even lower if combined with hyperthyroidism
1 DM, we compared IRRs within the type 1 DM group according
(IRR 0.65, 95% CI 0.51–0.85, p = 0.0012). Of the patients with
to the frequency of DKA, change in aDCSI score per year, and
type 1 DM complicated with hypothyroidism, the rates of live
daily insulin dose. All analyses were conducted using SAS sta-
births and fertility were similar to those without DM, but
tistical software (version 9.4; SAS Institute, Cary, NC).
higher than those without hypothyroidism.

3. Results
3.3. Live births, abortions, and fertility before and after a
diagnosis of type 1 diabetes (Table 3)
3.1. Characteristics of the study subjects (Table 1)

There were fewer live births before a diagnosis of diabetes

We identified 1191 female patients with type 1 DM aged
than after a diagnosis of diabetes (IRR 0.58, 95% CI 0.52–0.65
between 16 and 30 years in 2000, and 4764 age-, sex-, income-
vs. IRR 0.80, 95% CI 0.71–0.90). The number of abortions was
, and urbanization-matched control subjects from the NHIRD.
comparable between the two groups either before or after a
Fertility frequencies of 0, 1, 2, 3 in the DM and non-DM groups
diagnosis of DM. As a result, fertility was lower before a diag-
during the study period were 61.6%, 21.6%, 13.8%, and 2.9% vs.
nosis of diabetes than after a diagnosis of diabetes (IRR 0.68,
47.9%, 23.5%, 22.4%, and 6.3% respectively. Regarding comor-
95% CI 0.62–0.74 vs. IRR 0.87, 95% CI 0.79–0.95).
bidities, there were no significant differences in female infertil-
ity between the DM and non-DM groups. However, there were
3.4. The influence of diabetic ketoacidosis, aDCSI, and
significantly higher prevalence rates of hyperthyroidism and
daily insulin dose on live births, abortions, and fertility
hypothyroidism in the DM group than in the non-DM group
(Table 4)
(14.4% vs. 1.8% and 4.1% vs. 0.6%, respectively).

3.2. The influence of autoimmune thyroid diseases on live
births, abortions, and fertility during the study period
(Table 2)
During the study period, the DM group had a lower rate of live
births (IRR 0.67, 95% CI 0.62–0.73, p < 0.0001) than the non-DM
The rate of live births was similar in those with a higher fre-
quency of DKA; however, the abortion rate was significantly
increased. Fertility was proportional to the frequency of
DKA, although the difference did not reach statistical signifi-
cance. Patients with a change in aDCSI score of more than 0.5
per year had significantly lower live birth and fertility rates.
 154
diabetes research and clinical practice
Table 2 – The influence of autoimmune thyroid diseases on live births, abortions, and fertility during the study period.
Endpoints/Thyroid diseases DM (N = 1191) Non-DM (N = 4764) IRR 95% CI p-value
No. of Events PY IR No. of events PY IR

Live births 696 16383.23 4248.25 4214 66648.66 6322.71 0.67 0.62 0.73 <0.0001
Hyperthyroidism Yes 79 2050.48 3852.76 78 1091.89 7143.56 0.54 0.39 0.74 0.0001
No 617 14332.75 4304.83 4136 65556.76 6309.04 0.68 0.63 0.74 <0.0001
Hypothyroidism Yes 33 663.29 4975.23 25 391.96 6378.17 0.78 0.46 1.31 0.3488
No 663 15719.94 4217.57 4189 66256.70 6322.38 0.67 0.62 0.72 <0.0001
Abortions 393 16383.23 2398.80 1601 66648.66 2402.15 1.00 0.89 1.12 0.9802
Hyperthyroidism Yes 50 2050.48 2438.46 27 1091.89 2472.77 0.99 0.62 1.57 0.9533
No 343 14332.75 2393.12 1574 65556.76 2400.97 1.00 0.89 1.12 0.9562
Hypothyroidism Yes 16 663.29 2412.23 6 391.96 1530.76 1.58 0.62 4.03 0.3421
No 377 15719.94 2398.23 1595 66256.70 2407.30 1.00 0.89 1.11 0.9474

1 4 3 ( 2 0 1 8 ) 1 5 1 –1 5 8
Fertility 1089 16383.23 6647.04 5815 66648.66 8724.86 0.76 0.71 0.81 <0.0001
Hyperthyroidism Yes 129 2050.48 6291.22 105 1091.89 9616.32 0.65 0.51 0.85 0.0012
No 960 14332.75 6697.95 5710 65556.76 8710.01 0.77 0.72 0.82 <0.0001
Hypothyroidism Yes 49 663.29 7387.47 31 391.96 7908.94 0.93 0.60 1.46 0.7663
No 1040 15719.94 6615.80 5784 66256.70 8729.68 0.76 0.71 0.81 <0.0001
IR, incidence rate per 100,000 person-years; IRR, incidence rate ratio, comparing the type 1 diabetes cohort with the control cohort; PY, person-years.
 diabetes research and clinical practice 1 4 3 ( 2 0 1 8 ) 1 5 1 –1 5 8 155

Table 3 – Live births, abortions, and fertility before and after a diagnosis of type 1 diabetes.
Endpoints DM (N = 1191) Non-DM (N = 4764) IRR 95% CI p-value
No. of events PY IR No. of events PY IR

Live births
Before DM diagnosis 365 7328.18 4980.78 2512 29312.70 8569.66 0.58 0.52 0.65 <0.0001
After DM diagnosis 331 9055.05 3655.42 1702 37335.96 4558.61 0.80 0.71 0.90 0.0002
Abortions
Before DM diagnosis 194 7328.18 2647.32 782 29312.70 2667.79 0.99 0.85 1.16 0.9235
After DM diagnosis 199 9055.05 2197.67 819 37335.96 2193.60 1.00 0.86 1.17 0.9813
Fertility
Before DM diagnosis 559 7328.18 7628.09 3294 29312.70 11237.45 0.68 0.62 0.74 <0.0001
After DM diagnosis 530 9055.05 5853.09 2521 37335.96 6752.20 0.87 0.79 0.95 0.0028
IR, incidence rate per 100,000 person-years; IRR, incidence rate ratio, comparing the type 1 diabetes cohort with the control cohort; PY, person-
years.

The tertiles of daily insulin dose were less than 34.7 interna-
tional unit (IU), 34.7 to 46.9 IU, and more than 46.9 IU. A higher
daily insulin dose was significantly associated with higher
abortion and fertility rates.
4. Discussion
In this retrospective study on Taiwanese women of reproduc-
tive age with type 1 DM, we found lower live birth and fertility
rates in those with DM; however both rates were even lower
before a diagnosis of diabetes than after a diagnosis of dia-
betes. The abortion rate was comparable between the two
groups whether before or after a diagnosis of diabetes. This
is comparable to a previous study in Sweden [2], which
reported an overall 20% decrease in live births in female
patients who had already been diagnosed with type 1 dia-
betes. However, we assessed the frequency of giving birth,
not the actual number of children.
One detailed review article summarized the mechanism of
a decrease in fertility in females with type 1 DM [1]. First,
insulin insufficiency and fat loss induces low levels of leptin,
resulting in hypogonadotropic hypogonadism, delayed pub-
erty, and irregular menstrual cycle. Second, subcutaneous
insulin injections bypass hepatic clearance causing hyperin-
sulinemia, which contributes to increased androgen secretion
and the development of polycystic ovarian syndrome (PCOS).
Finally, hyperglycemia may cause glucose toxicity, and
advanced glycation end products may induce follicle apopto-
sis and cause early menopause. In this study, we confirmed
that the women with type 1 DM had a lower live birth rate
because of compromised gestation, not an increase in abor-
tion rate. In addition, type 1 DM, as a chronic and partly
genetic disorder, may have a negative effect on the patients’
intention to have children. Although we could not clarify
the use of contraception measures in our diabetic patients
and their spouses, no clinical guidelines suggested contracep-
tion during the study period. Moreover, we controlled for
urbanization and income, and the IRR was used to reduce
the influence of autonomous contraception due to reasons
other than type 1 diabetes.
The frequently overlap of type 1 DM and ATD is due to
common potential genetic susceptibility loci, such as human
leukocyte antigen (HLA) DQ and DR alleles, protein tyrosine
phosphatase, non-receptor type 22 (PTPN22) and cytotoxic T
lymphocyte-associated antigen-4 (CTLA-4) [5]. One multi-
center study found an increased prevalence of thyroperoxi-
dase and thyroglobulin antibodies in children and adoles-
cents with type 1 DM [15]. In addition, the prevalence of
autoimmune titers increased with increasing age, with the
highest in the 15- to 20-year age group (anti-
thyroperoxidase: 16.9%, anti-thyroglobulin: 12.8%). Hyperthy-
roidism is rare, accounting for only 0.46% of German and Aus-
trian adolescents with type 1 DM [16].
In a previous NHIRD study, patients with type 1 DM younger
than 18 years of age had a 6.6-fold higher IRR of ATD than nor-
mal subjects [17]. Interestingly, there was a high incidence rate
of thyrotoxicosis in both male and female patients. We
included older patients in the current study, and they have a
higher likelihood of having autoantibodies. In addition, we
used DDD of thyroid preparations instead of ICD-9-CM codes
to define ATD, which may have reduced the risk of miscoding
and excluded transient thyroiditis. However, we may have
underestimated the number of patients with subclinical thy-
roid diseases and those who chose radioiodine therapy or sur-
gical interventions. Nevertheless, we found similar IRRs, and
most of the patients had hyperthyroidism. Further studies
are needed to clarify this issue since hyperthyroidism is
related to glucose variability and the risk of DKA [16].
Women with ATDs have significantly higher levels of
endometrial T cells, which can lead to a hostile uterus. In
addition, vitamin D deficiency is more prevalent in women
with ATDs. Vitamin D has an anti-inflammatory effect and
it also activates the HOXA10 gene, which plays a critical role
in the implantation process. A low thyroid hormone concen-
tration can also induce ovarian dysfunction and an irregular
menstrual cycle. These mechanisms can result in decreased
fertility and pregnancy loss [6]. Menstrual irregularities are
common in patients with hyperthyroidism due to increased
levels of serum sex hormone binding globulin, testosterone,
and estradiol [18]. In the present study, the women with
hyperthyroidism and type 1 DM had a lower fertility rate,
which is consistent with previous studies. In addition, the
abortion rate was not significantly increased compared to
the control group, probably because patients with type 1 DM
routinely visit a hospital and because of the good accessibility
of medical services in Taiwan. We also found that the women
156 diabetes research and clinical practice 1 4 3 ( 2 0 1 8 ) 1 5 1 –1 5 8

Table 4 – The influence of diabetic ketoacidosis, aDCSI, and daily insulin dose on live births, abortions, and fertility.
Endpoints/DKA frequency DM (N = 1191) IRR 95% CI p-value
No. of events PY IR

Live births
DKA frequency
0–1 274 13712.40 1998.19 1 Reference
2–3 36 1618.40 2224.42 1.11 0.79 1.58 0.5452
>3 21 1052.42 1995.39 1.00 0.64 1.56 0.9951
Change in aDCSI score per year
<0.5 315 15030.35 2095.76 1 Reference
0.5 16 1352.88 1182.66 0.56 0.34 0.93 0.0256
Daily insulin dose (IU)
<34.7 94 5538.64 1697.17 1 Reference
34.7–46.9 123 5501.27 2235.85 1.32 1.01 1.72 0.0442
>46.9 114 5343.31 2133.51 1.26 0.96 1.65 0.1005

Abortions
DKA frequency
0–1 152 13712.40 1108.49 1 Reference
2–3 28 1618.40 1730.10 1.56 1.04 2.34 0.0304
>3 19 1052.42 1805.36 1.63 1.02 2.05 0.0450
Change in aDCSI score per year
<0.5 185 15030.35 1230.84 1 Reference
0.5 14 1352.88 1034.83 0.84 0.49 1.45 0.5315
Daily insulin dose (IU)
<34.7 50 5538.64 902.75 1 Reference
34.7–46.9 75 5501.27 1363.32 1.51 1.06 2.16 0.0240
>46.9 74 5343.31 1384.91 1.53 1.07 2.20 0.0194

Fertility
DKA frequency
0–1 426 13712.40 3106.68 1 Reference
2–3 64 1618.40 3954.52 1.27 0.98 1.66 0.0719
>3 40 1052.42 3800.75 1.22 0.88 1.69 0.2227
Change in aDCSI score per year
<0.5 500 15030.35 3326.60 1 Reference
0.5 30 1352.88 2217.49 0.67 0.46 0.96 0.0310
Daily insulin dose (IU)
<34.7 144 5538.64 2599.91 1 Reference
34.7–46.9 198 5501.27 3599.17 1.38 1.12 1.72 0.0030
>46.9 188 5343.31 3518.42 1.35 1.09 1.68 0.0063
aDCSI, adapted diabetes complication severity index; IR, incidence rate per 100,000 person-years; IRR, incidence rate ratio, compared with the
reference; IU, international unit; PY, person-years.

with type 1 diabetes with hypothyroidism had higher live
birth and fertility rates compared to those without hypothy-
roidism, probably due to adequate thyroxine replacement
therapy. Primary hypothyroidism is associated with elevated
thyrotropin-releasing hormone which also stimulates pro-
lactin secretion, and hyperprolactinemia and altered
gonadotropin-releasing hormone pulsatile secretion can then
lead to ovarian dysfunction [19]. Despite a lack of strong evi-
dence that thyroxine replacement can improve fertility, the
concentration of thyroid-stimulating hormone has been
inversely related to fertilization in women undergoing
in vitro fertilization [20]. In the 2017 American Thyroid Asso-
ciation guidelines, thyroxine replacement is recommended
for infertile women with overt hypothyroidism, as thyroid
function is reversible and thyroxine treatment is generally
safe and may exert a positive effect on fertility [21].
To the best of our knowledge, no previous study has inves-
tigated fertility in women with type 1 DM before they are
diagnosed with DM. As an autoimmune disease, type 1 DM
is classified into three stages: autoimmunity alone, autoim-
munity with dysglycemia, and symptomatic diabetes [22].
The 5-year and 10-year risks of symptomatic disease in stage
1 (autoimmunity alone) are about 44% and 70%, respectively.
Approximately 70% of patients with stage 2 (autoimmunity
with dysglycemia) develop symptomatic disease within 5
years. Thus, patients who have not yet been diagnosed with
DM may have asymptomatic dysglycemia for years, which
compromises fertility through the aforementioned mecha-
nisms. In addition, adequate insulin replacement and restora-
tion of thyroid function will be achieved soon after a
diagnosis under current treatment guidelines. This may
explain the better fertility rate after a diagnosis of DM.
 diabetes research and clinical practice
After a diagnosis of type 1 DM, the frequency of hospital-
izations for DKA is an indicator of glucose control status
and also a strong predictor of mortality [23]. Pregnancy can
be thought of as a state of accelerated starvation, since glu-
cose is readily absorbed by the fetus and placenta. Emesis
may also cause dehydration, insufficient glucose intake and
inadequate insulin replacement [24]. If DKA develops, osmo-
tic diuresis can decrease uteroplacental blood flow. In addi-
tion, maternal hyperkalemia and acidosis may cause
maternal arrhythmia, impaired placenta perfusion, and even-
tually fetal death [24]. In the current study, DKA significantly
increased the abortion rate but had no effect on live births.
Another reasonable explanation might be that the patients
with a higher number of abortions planned more gestations
to achieve the number of live births they wanted.
In the present study, a greater increase in aDCSI score per
year represented worse glucose control status and more com-
plications, and resulted in a lower live birth rate. As men-
tioned, insulin insufficiency and glucose toxicity can cause
delayed menarche, irregular menstrual cycle, and early
menopause, which shorten the reproductive years [1]. More-
over, a higher daily insulin dose was significantly associated
with a higher abortion rate. Although body weight data are
not available in the NHIRD, a higher daily insulin dose may
be associated with either insulin resistance or obesity, which
are features of PCOS. Moreover, 24% of patients with type 1
diabetes have been reported to have PCOS, mainly due to
iatrogenic hyperinsulinemia [25]. Unfortunately, pregnancy
outcome data are lacking in this specific group of patients.
However, in the general PCOS population, the rate of early
pregnancy loss (first trimester miscarriage) has been reported
to be higher (30–50%) compared to normal women (10–15%)
[26]. Hyperandrogenemia, insulin resistance, impaired fibri-
nolysis, and endometrial dysfunction may explain the high
early pregnancy loss rate. However, these etiologies remain
unclear in patients with type 1 diabetes with PCOS, and fur-
ther studies are needed to elucidate this issue.
Due to the retrospective database design, there are some
limitations to this study. First, the NHIRD does not contain
serological data, and therefore we could not evaluate thyroid
autoimmune titers or the severity of ATDs. In addition, we
could not clarify the influence of thyroid function in the
patients with hyperthyroidism who received surgical or radio-
iodine ablation. Second, detection bias probably existed, since
patients with type 1 DM are suggested to undergo routine thy-
roid function checks soon after a diagnosis. The IRR of ATD
may therefore be overestimated. Third, marriage status and
educational level are not available in the NHIRD. Neverthe-
less, marriage status is not a necessary condition for fertility,
and we matched urbanization and income, which are highly
associated with education level, between the DM and non-
DM groups. Despite these caveats, the strengths of this study
include its nationwide, population-based cohort design, long
duration of follow-up and large number of patients with type
1 diabetes.
In conclusion, this pilot study demonstrated that type 1
DM compromises female fertility, even before it is diagnosed.
Associated hyperthyroidism further decreases fertility; how-
ever hypothyroidism with adequate treatment may improve
fertility. Diabetic ketoacidosis and a higher daily insulin dose
1 4 3 ( 2 0 1 8 ) 1 5 1 –1 5 8 157
were strongly associated with pregnancy loss. Diabetic com-
plications significantly reduced the number of live births.
Our research supports the benefits of routine thyroid function
surveys in patients with type 1 diabetes to improve the fertil-
ity rate. Blood glucose and thyroid function surveillance in
infertile females may allow for an early diagnosis of type 1
diabetes and associated thyroid disease.
Acknowledgements
The author’s would like to thank the Health Information and
Epidemiology Laboratory of Chang Gung Memorial Hospital,
Chiayi Branch (CLRPG6G0041) for their comments and assis-
tance in data analysis. This study was supported by a grant
from Chang Gung Memorial Hospital, Chiayi Branch, and
based on the National Health Insurance Research Database
provided by the Central Bureau of National Health Insurance,
the Department of Health, and managed by the National
Health Research Institutes. The interpretation and conclu-
sions contained herein do not represent those of the Bureau
of National Health Insurance, Department of Health, or
National Health Research Institutes.
Authors’ contributions
Y-HL designed the study and drafted the manuscript. K-JC col-
lected, arranged, filtered, and managed data. Y-SP drafted the
manuscript. Y-HY drafted the manuscript and contributed to
the statistical analysis. P-CC assisted with the study design
and revised the content. Y-HY is the guarantor of this work
and, as such, had full access to all the data in the study and
takes responsibility for the integrity of the data and the accu-
racy of the data analysis. All authors revised the manuscript.
Conflict of interest
None.
Funding
This research did not receive any specific grant from funding
agencies in the public, commercial, or not-for-profit sectors.
Appendix A. Supplementary material
Supplementary data associated with this article can be found,
in the online version, at https://doi.org/10.1016/j.diabres.2018.
07.010.
R E F E R E N C E S
[1] Codner E, Merino PM, Tena-Sempere M. Female reproduction
and type 1 diabetes: from mechanisms to clinical findings.
Hum Reprod Update 2012;18:568–85.
[2] Jonasson JM, Brismar K, Sparén P, Lambe M, Nyrén O,
Ostenson CG, et al. Fertility in women with type 1 diabetes.
Diabetes Care 2007;30:2271–6.
158 diabetes research and clinical practice
[3] Sjöberg L, Pitkäniemi J, Haapala L, Kaaja R, Tuomilehto J.
Fertility in people with childhood-onset type 1 diabetes.
Diabetologia 2013;56:78–81. [16]
[4] Wiebe JC, Santana A, Medina-Rodrı́guez N, Hernández M,
Nóvoa J, Mauricio D, et al. Fertility is reduced in women and
in men with type 1 diabetes: results from the Type 1 Diabetes
Genetics Consortium (T1DGC). Diabetologia 2014;57:2501–4. [17]
[5] Barker JM. Clinical review: Type 1 diabetes-associated
autoimmunity: natural history, genetic associations, and
screening. J Clin Endocrinol Metab 2006;91:1210–7.
[6] Twig G, Shina A, Amital H, Shoenfeld Y. Pathogenesis of [18]
infertility and recurrent pregnancy loss in thyroid
autoimmunity. J Autoimmun 2012;38:J275–81.
[7] Lin WH, Wang MC, Wang WM, Yang DC, Lam CF, Roan JN,
et al. Incidence of and mortality from Type I diabetes in [19]
Taiwan from 1999 through 2010: a nationwide cohort study.
PLoS One 2014;9:e86172. [20]
[8] National Health Insurance Research Database, Taiwan.
<http://nhird.nhri.org.tw/en/index.htm> [Accessed 03 March
2017].
[9] Chou IC, Wang CH, Lin WD, Tsai FJ, Lin CC, Kao CH. Risk of [21]
epilepsy in type 1 diabetes mellitus: a population-based
cohort study. Diabetologia 2016;59:1196–203.
[10] Pan ML, Chen LR, Tsao HM, Chen KH. Relationship between
polycystic ovarian syndrome and subsequent gestational
diabetes mellitus: a nationwide population-based study. PLoS [22]
One 2015;10:e0140544.
[11] Ko KY, Yen RF, Lin CL, Cheng MF, Huang WS, Kao CH.
Pregnancy outcome after I-131 therapy for patients with
thyroid cancer. Med (Baltimore) 2016;95:e2685.
[12] WHO Collaborating Centre for Drug Statistics Methodology. [23]
<https://www.whocc.no/> [Accessed 30 April 2017].
[13] Chang HY, Weiner JP, Richards TM, Bleich SN, Segal JB.
Validating the adapted diabetes complications severity index [24]
in claims data. Am J Manag Care 2012;18:721–6.
[14] Young BA, Lin E, Von Korff M, Simon G, Ciechanowski P, [25]
Ludman EJ, et al. Diabetes complications severity index and
risk of mortality, hospitalization, and healthcare utilization.
Am J Manag Care 2008;14:15–23. [26]
[15] Kordonouri O, Klinghammer A, Lang EB, Grüters-Kieslich A,
Grabert M, Holl RW. Thyroid autoimmunity in children and
1 4 3 ( 2 0 1 8 ) 1 5 1 –1 5 8
adolescents with type 1 diabetes: a multicenter survey.
Diabetes Care 2002;25:1346–50.
Dost A, Rohrer TR, Fröhlich-Reiterer E, Bollow E, Karges B,
Böckmann A, et al. Hyperthyroidism in 276 children and
adolescents with type 1 diabetes from Germany and Austria.
Horm Res Paediatr 2015;84:190–8.
Lu MC, Chang SC, Huang KY, Koo M, Lai NS. Higher risk of
thyroid disorders in young patients with Type 1 diabetes: a
12-year nationwide, population-based retrospective Cohort
study. PLoS One 2016;11:e0152168.
Mintziori G, Kita M, Duntas L, Goulis DG. Consequences of
hyperthyroidism in male and female fertility:
pathophysiology and current management. J Endocrinol
Invest 2016;39:849–53.
Poppe K, Velkeniers B, Glinoer D. Thyroid disease and female
reproduction. Clin Endocrinol (Oxf) 2007;66:309–21.
Cramer DW, Sluss PM, Powers RD, McShane P, Ginsburgs ES,
Hornstein MD, et al. Serum prolactin and TSH in an in vitro
fertilization population: is there a link between fertilization
and thyroid function? J Assist Reprod Genet 2003;20:210–5.
Alexander EK, Pearce EN, Brent GA, Brown RS, Chen H,
Dosiou C, et al. Guidelines of the American Thyroid
Association for the diagnosis and management of thyroid
disease during pregnancy and the postpartum. Thyroid
2017;27:315–89.
Insel RA, Dunne JL, Atkinson MA, Chiang JL, Dabelea D,
Gottlieb PA, et al. Staging presymptomatic type 1 diabetes: a
scientific statement of JDRF, the endocrine society, and the
american diabetes association. Diabetes Care
2015;38:1964–74.
Gibb FW, Teoh WL, Graham J, Lockman KA. Risk of death
following admission to a UK hospital with diabetic
ketoacidosis. Diabetologia 2016;59:2082–7.
Parker JA, Conway DL. Diabetic ketoacidosis in pregnancy.
Obstet Gynecol Clin North Am 2007;34:533–43.
Escobar-Morreale HF, Roldán-Martı́n MB. Type 1 diabetes and
polycystic ovary syndrome: systematic review and meta-
analysis. Diabetes Care 2016;39:639–48.
Kamalanathan S, Sahoo JP, Sathyapalan T. Pregnancy in
polycystic ovary syndrome. Ind J Endocrinol Metab
2013;17:37–43.