Maternal Factors Affecting Pregnancy Outcomes in Hypothyroid Women of Pakistan: A Retrospective Study From a Tertiary Center. (MHPO-II)

Background: Hypothyroidism in pregnancy is known to be associated with adverse outcomes depending on geographical distribution and diagnostic criteria. We analyze the maternal factors affecting pregnancy outcomes in maternal hypothyroidism for the �rst time in Pakistan. Methods: We conducted a retrospective study on pregnant women attending obstetric clinics during the year 2008-2016. We noted maternal factors such as maternal age, gestational age at antenatal visit, gestational diabetes mellitus, hypertension, and past obstetric history of miscarriages affecting pregnancy loss or live birth in pregnant hypothyroid women as well as normal healthy pregnant women. Odds ratio was calculated using multivariate logistic regression. Subgroup analysis was carried out to compare women with overt versus subclinical hypothyroidism. All data analysis was performed on Stata version 12. Results: Maternal age, gestational age at antenatal visit, gestational diabetes mellitus, hypertension, and past obstetric history of miscarriages were signi�cantly associated with the outcome at the univariate analysis. At multivariable analysis, maternal age, gestational age at the antenatal visit, and past obstetric history of miscarriages were associated with an increased likelihood of pregnancy loss, whereas, gestational diabetes mellitus and hypothyroidism were found to be counter-protective. Subgroup analysis revealed no difference between women those with overt versus subclinical hypothyroidism. Conclusion: We report maternal factors such as age, gestational age at the antenatal visit, and past obstetric history of miscarriages as signi�cant factors affecting adverse pregnancy outcomes in hypothyroidism for the �rst time from Pakistan.


Background
Hypothyroidism is one of the most common endocrine disorders in women of reproductive age (1).During pregnancy, it is associated with complications like gestational hypertension (GH) (2-7) and fetal death or abortions (8,9).Further, hypothyroidism is also known to be related to preterm birth, placental abruption, and stillbirth (10)(11)(12)(13)(14)(15)(16)(17)(18) .Euthyroid women with autoimmune thyroid diseases also show impairment of thyroid function during gestation and suffer from a high rate of obstetric complications (19).Therefore, hypothyroidism is a high-risk medical disorder in pregnancy.
Data from the South Asian region about maternal factors affecting pregnancy outcomes in hypothyroid women is limited.An Indian study showed 15.4% intrauterine growth retardation and 3.9% perinatal mortality among 26 pregnancies affected by hypothyroidism (20).Another study reported that overt hypothyroid women were more prone to have GH (p = 0.04), intrauterine growth restriction (IUGR) (p = 0.01) and intrauterine fetal death (p = 0.0004) (21).A study from Pakistan conducted on patients with gestational diabetes depicted 61.5% subclinical hypothyroidism.In this study, thyroid-stimulating hormone (TSH) remained independently associated with random blood glucose (r = 0.109; p < 0.05), poor glycemic control (r = 0.227; p < 0.001) and negatively associated with fetal growth (r = − 0.206; p < 0.001) (22).Another study reported an association between thyroid dysfunction and infertility (23).
Despite signi cant evidence of hypothyroid related maternal and neonatal complications, universal screening is a matter of debate among international societies.Several studies are reporting obstetric outcomes in both overt and subclinical hypothyroidism, but none of them have a common endpoint (24,25).The main reason being lack of uniformity across population-based studies, as thyroid functions are affected by dietary iodine intake, environmental factors, genetic predisposition, and prevalence of the autoimmune phenomenon.Therefore, it is important to evaluate individual population prevalence and maternal factors affecting pregnancy outcomes of hypothyroid patients before embarking on universal recommendations.
In this study we aim to analyze retrospectively through medical records at Aga Khan University Hospital (AKUH), the maternal factors affecting pregnancy outcomes in hypothyroid mothers, whether diagnosed before conception or during pregnancy.To the best of our knowledge, this is the rst study from Pakistan.

Data Collection:
We conducted a retrospective data collection on pregnant patients attending either endocrine or obstetric clinics at our hospital during the year 2008-2016.We divided the data into a study group, including those patients with hypothyroidism and the control group with normal pregnancies without any comorbidities.
Trained medical o cers collected the data and the principal investigator randomly cross-checked and veri ed it.Data from maternal les included maternal age and gestational age (trimesters) at their rst visit.History of pre-gestational diabetes (DM), gestational diabetes (GDM), and hypertension [Chronic Hypertension (HTN), gestational hypertension (GH), and preeclampsia] were noted as comorbid factors in the study group.We also recorded past obstetric history for miscarriages in both groups.TSH levels were also recorded throughout the gestational period.Pregnancy outcomes were divided into live births and pregnancy loss.Pregnancy loss included; abortion, stillbirths and intrauterine death, medical termination of pregnancy, and ectopic pregnancy.

Patient selection:
Inclusion criteria: Study group included all pregnant women attending endocrine and obstetric clinics who had been diagnosed either 'overt hypothyroidism' or 'subclinical hypothyroidism' or simply labeled 'hypothyroidism' by the clinician either before or during pregnancy.The control group was pregnant women without any comorbidities attending the obstetric clinic in the same period.
Exclusion Criteria: Pregnant women with any abnormal thyroid function tests that did not t into the criteria of overt or subclinical hypothyroidism and those not labeled were excluded from the study.

Pregnancy Outcomes:
We de ned pregnancy outcomes as either live birth or pregnancy loss in our study.Pregnancy loss included Abortion; de ned as loss of a pregnancy at or before 20 weeks of gestation, intrauterine death (IUD) or stillbirth; de ned as a pregnancy loss occurring after 20 weeks of gestation (26,27).Other outcomes include Medical termination of pregnancy; de ned as the termination of pregnancy in any trimester through medical procedures according to the clinical guidelines of the Royal College of Obstetrics and Gynaecology (RCOG) (28).And, Ectopic pregnancy; which occurs when the developing blastocyst implants at a site other than the endometrium of the uterine cavity, as determined by clinical and radiological examinations (29).

Thyroid Stimulating hormone levels:
TSH levels available before conception and during each month throughout pregnancy were noted.TSH levels were assessed by Advia Centaur (Siemens Diagnostics), Chemiluminescence immunoassay.The functional sensitivity limit of the assay is 0.008 µIU/mL, with the assay detection range spanning from 0.008-150 µIU/mL to detect the lowest and highest abnormal value.The normal laboratory reference range of TSH for adults is 0.4-4.2µIU/mL for the age group 21-54 years in our hospital.

Statistical Analysis:
Descriptive Statistics: We reported frequencies with percentages for all the independent variables for the study as well as control groups as shown in Table 1.Binary logistic regression analysis technique was used in this study, to assess the association of independent variables with the pregnancy outcomes (dependent variable).First, we applied the technique of univariate logistic regression analysis and our p-value cut-off was taken at 0.25, which meant that the independent variable has a p-value of 0.25 or less and was considered statistically signi cant.In the univariate regression model, we put in all the independent variables one by one (on an individual basis) that we had in the dataset and all were found to be signi cant, according to the cut-off set at this level.
For multivariable logistic regression analysis, all the signi cant variables that were identi ed at the univariate analysis level were used.Variable having the lowest p-value was entered in the multivariable model rst, followed by the higher p-value, and the variable with the highest p-value was entered in the last, following a manual model building approach.Five variables were found statistically signi cant at this level, having a p-value cut-off of less than or equal to 0.05.

Results
Table 1 shows the classi cation of independent variables in two groups that include pregnant women with hypothyroidism (N = 708) and normal pregnant women without any comorbidities (N = 759).The mean age in pregnant women with hypothyroidism is 30.91 (± 5.68) while the mean age of normal pregnant women is 29.44 (± 4.85).In both groups, the majority of the pregnant women had presented in their rst trimester (study group; 53.8%, control group; 44.4%) and at least 86.3% of the study group had developed hypothyroidism before pregnancy.The top three comorbidities of pregnant women in the study group were; gestational diabetes mellitus (21.2%), gestational hypertension (10.6%), and pre-gestational diabetes (7.8%).

Maternal factors affecting outcomes:
Since our outcome was a binary variable, we applied binary logistic regression, taking pregnancy loss as a positive outcome and giving it a code of 1.On the other hand, live births were coded as 0 representing no outcome being reported.Table 2 shows the results of univariate logistic regression analysis.All independent factors such as maternal age (p-value = 0.01), gestational age at the antenatal visit (p-value < 0.01), gestational diabetes mellitus (p-value < 0.01), hypertension (p-value = 0.12), and past obstetric history of miscarriages (p-value < 0.01) were signi cantly associated with the pregnancy outcome at the univariate analysis level.Upon multivariable analysis (as shown in Table 3), woman who falls in the age group of 31 to 40 years had 1.55 times more chances of having a pregnancy loss as compared to women belong to the age group of 19 to 30 years.Similarly, women who had a past obstetric history of miscarriages were 3.11 times more likely to have a pregnancy loss compared to those who did not.According to our data, pregnant women with hypothyroidism are 36% less likely of having a pregnancy loss as compared to normal pregnant women.Moreover, pregnant mothers who had GDM were 96% less likely to have a pregnancy loss compared to those who didn't have gestational diabetes.Lastly, pregnant women who presented in their third trimesters were 81% less likely to have pregnancy loss compared to those who presented in their rst trimester.
Comparison between women diagnosed with (overt and subclinical) hypothyroidism either during pregnancy or before pregnancy: At least 86.3% of the study group had developed (overt and subclinical) hypothyroidism before pregnancy.The outcome was compared among women who were diagnosed with hypothyroidism during pregnancy and women who were diagnosed with hypothyroidism before pregnancy using chi-square.The worse outcome was not statistically signi cant in pregnant women diagnosed with hypothyroidism during pregnancy as compared to women who were diagnosed with hypothyroidism before pregnancy (pvalue = 0.628).
Subgroup analysis of pregnant women with overt hypothyroidism and subclinical hypothyroidism: Table 4 shows independent maternal factors based on type of hypothyroidism in study group.Table 5 shows the results of subgroup analysis when women with overt and subclinical hypothyroidism respectively are analysed separately.Results show that there was no signi cant difference between maternal factors affecting pregnancy outcomes in subclinical hypothyroid patients as compared to those with overt hypothyroidism.The only exception was gestational diabetes mellitus, which is signi cantly associated with pregnancy outcome (p-value = 0.002) in only subclinical hypothyroid patients.There was no case of GDM in overt hypothyroid cases.

Discussion
Our study found maternal age, gestational age at antenatal visit, and history of miscarriages, as major maternal factors signi cantly affecting pregnancy outcomes in hypothyroid mothers, whereas the presence of gestational diabetes and hypothyroidism had a favourable effect.
There is su cient evidence to support the effect of increasing maternal age on obstetric outcomes.In a study of 534 pregnant women, there was an independent association of risk of miscarriage among women older than 35 years of age (7.7%), although only 16 patients had increased TSH (30).Therefore, an older hypothyroid pregnant woman is at higher risk of adverse obstetric outcomes than a younger woman as also evident our ndings.
Data is scarce on the effect of the previous history of miscarriages an independent factor for pregnancy loss or adverse pregnancy outcome in hypothyroid women.The subsequent pregnancy rate did not differ between the subclinical and euthyroid pregnant groups (55.4%, 31/56 vs 51.3%, 134/261, respectively) in a retrospective study, although the pregnancy loss rate (< 22 weeks of gestation) tended to be higher in the borderline-subclinical hypothyroid than the euthyroid group (29.0%, 9/31 vs 17.9%, 24/134; P = 0.16) (31).Similar results were reported in another study including those treated or untreated (32).However, there is plenty of data on the association of recurrent miscarriages and hypothyroidism (33).We need to analyse the ndings in our study with prospective study designs in the future.
There is a signi cant association of hypothyroidism as a maternal factor affecting pregnancy outcomes in our study population.Several studies report an adverse effect on the fetal outcome due to the presence of thyroid disorders, especially hypothyroidism (21,34).On the other hand, data also contradict this nding with studies describing no signi cant effect on pregnancy loss when compared with normal pregnant women (35,36).Although, our study also did not support these ndings, however, being diagnosed as hypothyroid was in fact, counter-protective.TSH levels remained in nearly euthyroid range in our population, as in most cases they were timely initiated on treatment.However, we do not have our trimester-speci c ranges of TSH, therefore the clinical signi cance of these cut-offs is under the guidance of ATA.There was no difference in the outcomes of whether our women were diagnosed with hypothyroidism before or during pregnancy.There was also no difference in the outcomes of women whether they have overt or subclinical hypothyroidism in our population.The only exception was related to GDM, in women with subclinical hypothyroidism, as we have no GDM in overt hypothyroidism women.Carefully, analysing these ndings further in large multicentre prospective trials is required to assess if this is because of our special obstetric care in such patients and other patient-based factors, or as they were timely started on levothyroxine treatment.
In a Chinese study, comparing the effects of different diagnostic criteria of subclinical hypothyroidism on pregnancy outcomes, the incidence of miscarriage, premature delivery, gestational hypertension and gestational diabetes mellitus (GDM) in the American Thyroid Association (ATA) based groups had statistically signi cant differences (p-value of < 0.05) when compared with the control group (2,37).Two decades back, Gonzalez et al. showed that 26% of pregnant women with type 1 diabetes and 4% of healthy pregnant women had thyroid dysfunction.(38) Similarly, in another study by Ortega-Gonzalez et al, there were 50 healthy pregnant women, 50 GDM women and 50 pregnant women with type 2 diabetes, the thyroid peroxidase antibodies (TPO Ab) > 251 U/ml ( strongly positive) was found in 10% of healthy women, 10% of pregnant women with type 2 diabetes and 6% in women with GDM without statistically signi cant difference (39).One of the recent studies, however, described that the rate of thyroid dysfunction in GDM patients is similar to normal pregnant control women (40).The presence of these dual endocrinopathies is also analysed in pregnant women for several adverse pregnancy-related outcomes, with one of the studies reporting a signi cant rate of spontaneous rst trimester abortions (9.3%) as compared to either comorbidity alone (41,42).Interestingly, most studies are redundant to describe the independent association of these maternal factors with pregnancy loss.Moreover, studies are assessing the risk of developing GDM in hypothyroid pregnancies, rather than evaluating GDM as an independent factor affecting pregnancy outcomes in hypothyroid pregnant women, which is described in our study (43)(44)(45).Our data reported a signi cant protective effect of the presence of GDM in hypothyroid women (subclinical cases) with respect to pregnancy outcome.We assume this difference may be due to the presence of a separate health care system established for patients with GDM in our hospital, as we have a combined maternal diabetes clinic setup including multidisciplinary experts dealing with these groups of patients.
A limited amount of data is present on the timing of the rst visit (the gestational age) at the healthcare centre by hypothyroid pregnant women.Jahan et al. however, did not nd a statistically signi cant association between rst-trimester pregnancy loss and high normal TSH level (Pearson Chi-square = 2.01, p = 0.1558) from Bangladesh (46).However, in our study, many patients were diagnosed with hypothyroidism before conception and were already on levothyroxine replacement.Although, majority of them did not have their preconception TSH levels measured (which is also our limitation due to retrospective study design), still those who presented early (second trimester) were more likely to have pregnancy loss compared to those who presented late (third trimester).
Our study had several limitations.Firstly, it was a retrospective study design due to which we could not compare all maternal characteristics between the study and control groups.Secondly, hypothyroidism and GDM as comorbidities providing favourable effect, need further analysis based on large randomized control trials, which in itself remains an ethical limitation in the obstetric population.However, this is the rst study of its kind from Pakistan, and we had taken a good sample size for analysis and interpretation.Further, we recommend prospective cohort and case-control studies to assess risk factors for pregnancy loss catering to a large representative sample in our region.

Conclusion
We report the maternal factors affecting pregnancy outcomes for the rst time from the region of Pakistan.Maternal age, gestational age at antenatal visit, and previous history of miscarriages are found to be signi cantly affecting adverse pregnancy outcomes in our study, however, gestational diabetes and hypothyroidism had a protective effect on our population.

Table 1
Independent maternal factors of the pregnant population.

Table 2
Univariate analysis of independent maternal factors affecting pregnancy outcomes in both groups.
0.05 was taken as statistically signi cant at the level.Variable of maternal age, gestational age at the antenatal visit, past obstetrical history of miscarriages, gestational diabetes mellitus, and hypothyroidism were signi cantly associated with pregnancy outcome at the multivariable analysis level.

Table 3
Multivariable analysis of maternal factors affecting pregnancy outcomes in both groups.

Table 4
Independent maternal factors based on type of hypothyroidism in study group.

Table 5
Subgroup analysis of maternal factors affecting pregnancy outcomes based on type of hypothyroidism (either before or during pregnancy).