Feasibility of setting up a birth cohort and screening Gestational Hyperglycemia in Public health facilities in Bengaluru, India

Introduction

India is locale to nearly 62 million individuals with Type 2 Diabetes Mellitus (T2DM) and is projected to increase to 79.4 million by 2030.^1,2 Gestational Hyperglycemia (GH) is one of the predisposing factors for the onset of T2DM in women. It is estimated that approximately 1.5 to 17% of Indian pregnant women may have hyperglycemia. GH may also induce changes in the intrauterine environment resulting in adverse health consequences in offspring such as increased risk of childhood obesity and premature onset of T2DM. Among the known methods for assessing glucose levels in pregnancy, the random and fasting glucose tests are relatively ineffective for screening higher glucose levels in pregnancy due to their instability and laboratory variability. The public health facilities in India sparingly use Oral Glucose Tolerance Test (OGTT), the gold standard test for screening GH. Apart from this, there are several other challenges. These include low awareness regarding GH screening in pregnant women and health professionals, lack of standard operating procedures for screening and inadequate care and attention provided to pregnant women.^3,4 There is limited evidence in exploring the use of OGTT as a screening test in public health facilities and monitor the adverse consequences of GH in Indian women and children. Despite using some notional method at most public health centres, the current screening technique and methods are of limited clinical utility.³ Only prospective cohort studies will help in understanding how the higher levels of glucose levels in mothers may affect childhood obesity, a marker for risk of T2DM. Birth cohort recruits a group of infants born during a similar time frame prospectively following -them over a period of time., serving as a crucial research resource in the scientific landscape of any country.^5,6 Suchstudies are vital to determine pregnancy related risks and neonatal outcomes.^5,7 These cohorts are also helpful in studying several other influences of maternal and early childhood environment on disease risk in adolescence and adult life^5-9. 48.2%of the pregnant women visit public health facilities in India¹⁰. However, all the birth cohort studies in India⁴ have been tried either in private settings or in general population,¹¹ with none of them in public health facilities⁵ . Till date, no study has explored whether it is possible to start and build birth cohorts in the Indian public health facilities.^12,13 Designed as a pilot for recruiting a larger cohort, the aim of this study was to explore the feasibility of recruiting and administering Oral glucose tolerance tests (OGTT) in pregnant women attending antenatal clinics at public health facilities in Bengaluru city. The study also aimed to ascertain the logistics, infrastructure and support system required to conduct and carry out a similar designed study for a longer period of time.

Materials and methods

Study Setting: We explored the feasibility of using public health facilities in the jurisdiction of the city corporation of Bengaluru, the Bruhat Bengaluru MahanagaraPalike (BBMP). Accordingly, we obtained approval for setting up birth cohort at three maternity hospitals namely Banashankari Referral hospital and H Siddhaiah Road referral Hospital. The protocol of this study has been published earlier¹³. Pregnant women visiting the antenatal clinics at the selected health facilities.

Eligibility criteria: Pregnant women in the age group of 18- 40 years with over 24 weeks of gestation planning to reside in the study area for the next one year and willing to provide informed consent were included in the study. For the feasibility phase, we excluded women with coexisting severe illness, treated previously for infertility, twin pregnancy, and those with more than 4 live births. The feasibility parameters include recruitment, administration of questionnaire, anthropometry, conduct of oral glucose tolerance test, transport of samples to the reference lab and follow up of respondents.

Screening & recruitment: The enrollment phase began from the month of Nov 2013 at one public health hospital (Banashankari referral hospital). This was expanded to another centre in July 2014. (H Siddaiah Road Referral Hospital) All the women attending antennal clinic at the referral hospital were screened and their contact information was noted.  

The feasibility of obtaining baseline information was done using a paper-based questionnaire, administered by the research assistants. This questionnaire sought information on socioeconomic profile, medical history, 24-hour diet recall, food frequency questionnaire, social support, Kessler’s depression scale and physical activity. We also explored measuring the weight, height, and skin fold thickness of biceps, triceps and sub-scapular region of infants using Holtain calipers.

OGTT Administration and sample collection: The enrolled pregnant women were referred to undergo for Oral Glucose Tolerance tests. Two samples were obtained from each respondent: one during fasting and the other at two hours after consumption of 75gm of glucose solution. In order to prevent in- vitro glycolysis, care was taken to centrifuge, refrigerate and transport the samples within one hour so that the integrity of the samples was maintained. Blood was collected in a lithium-heparin tube and centrifuged. The tubes were marked with respondent ID, name, and type of sample (fasting, postprandial). A log sheet was maintained for each sample drawn. The samples were then transported to the laboratory in ice packs. Glucose concentration was determined by Hexokinase analysis. The centralized laboratory would share the results on the same day of testing. If the glucose values were in normal range, the lab reports were handed over to the respondent at their next ANC visit; otherwise were contacted immediately, advised on healthy lifestyle and were referred to the gynecologist present at the health center.

Follow-ups: The feasibility of conducting followup visits of mother and child were planned at birth and concurring with scheduled immunized visits of children. Generally, there are four immunizations scheduled for the child in its first year of life- at 6th week, 10th week, 14th week and at 9th month. At each time of follow-up, we attempted to obtain anthropometric measurements of both mother and child. The recruitments and follow-ups progressed simultaneously at both the study centers until the month of June 2015.

Anthropometric measurements: The measurements were taken for all the women and children using of standard techniques¹⁴. In order to assess obesity, the feasibility of obtaining the measurements was explored. The infant's weight, length, head, chest, waist and hip circumference were measured. The skin fold-thickness measurements where measured with use of a Holtain Tanner/Whitehouse skin fold caliper; (Holtain Ltd, Crosswell, Crymych, Pembs., SA41 3UF, UK. Registered in Wales No. 646316). The calipers have been specifically designed for the accurate measurement of subcutaneous tissue; it was developed in close collaboration with the London University Institute of Child Health. It incorporates the recommended principles for standard usage in such measurements at the triceps, biceps and subscapular sites. Trained Research Assistants made all measurements, and the inter-observer variation for the skin-fold measurement ranged between 0.1 mm to 0.9 mm.

Data analysis: The collected information was compiled and entered in Microsoft access database and analyzed using SPSS (statistical package of social science) version 20. The recruitment was assessed by analyzing proportion of people enrolled after screening and reasons for exclusion. The ease of administering the OGTT was assessed from the logistic point of view and questionnaire administration was assessed qualitatively. The follow-ups were assessed based on the proportion of completion of follow-ups at each scheduled visit.

Ethical Considerations: The identity of the participants was kept anonymous and no names or other personal identifying information were retained in the final dataset. The study was reviewed, and ethical approval was obtained from Institutional Ethics Committee, Indian Institute of Public Health-Hyderabad, (IIPH H), Bengaluru campus. Written informed consent was obtained from all the participants (Web Appendix-1). Administrative approval was taken from the office of the chief health officer of BBMP.

Results

Screening

We attempted in a total of 1243 women visiting the ANC at both study sites for enrolling in the research study. Among them, we excluded 279 (22%) for not meeting the eligibility criteria. The reasons for exclusion of pregnant women are listed in table 1. These include, on-residents of study area and those who had planned to deliver at a different hospital (119), women who had received treatment for infertility (32), women who were not interested in participating (31), thyroid disorder (26), not available for follow up visits as immunization would be done at native place (21) asthma(10), language barrier(8), hypertension(7), epilepsy(7), few women had already undergone OGTT elsewhere and did not see any benefit in joining the study (7), women with twin pregnancy (6), diabetes mellitus (3), less than 18 year old (1), women who have had more than four live births(1).

Feasibility of Recruitment and Questionnaire administration

Among the screened women, Informed consent was obtained from 964 respondents. Of the enrolled women, 63 (7%) did not complete the questionnaire and anthropometry. (Figure-1) Based on the field-testing, the average time taken for administering the questionnaire was approximately 45 minutes. It was challenging to obtain certain sensitive information from the respondents like husband's smoking and drinking habits. The feedback from research assistant was that questionnaire was very extensive. Particularly, the food frequency questionnaire over the past one-month took up most of the time as respondents were unable to recall and give definite answers. The respondents also had difficulty in answering the physical activity questionnaire wherein they had to specify the time they spent on their daily activities. As a result, the respondents would lose interest in the middle of the interview or were persuaded by their family members to quit the study interview midway as they have other impending tasks to attend to. The study team made midcourse corrections and the questionnaire was divided into sections, which could be administered in two or three sittings corresponding to the visits of respondents to the hospital. In addition, due to cultural inhibitions, respondents were very shy to undergo anthropometry measurements.

Data entry

Following the administration of the questionnaire by research assistant, all the data collected in the paper form were entered into a MS Access database specifically created for this study. Two databases were created in access; one for baseline forms and one for follow up forms. This would take approximately 30 minutes depending on proficiency. Therefore, the total amount taken to administer and enter the data was around nearly 75 minutes per participant.

Feasibility of OGTT administration

After the questionnaire was administered on the day of ANC visit, each woman would be scheduled for OGTT and was asked to visit the health centre after overnight fast. Among the enrolled women, 864(90%) underwent OGTT while 100(10%) did not turn up for OGTT due to varying reasons, Including lack of interest (40), shifted to another town or their native (37), some were untraceable (10), few had got the OGTT done at private hospital and were not interested in enrolling into the research study (5), preterm delivery (4), few expressed fear for undergoing blood test (4) and some were not willing to fast for the lab test(3). Among the untraced pregnant women, the reasons included not answering calls, providing wrong contact number or had no number at all. While administering OGTT, we found that 32 women (3.7%) experienced discomfort in the form of nausea, giddiness, fainting. The prerequisites of an overnight fast and rapidly drinking an extremely sweet solution on an empty stomach might trigger the symptoms of morning sickness. Postprandial sample was not drawn in those women who vomited and their test was rescheduled. Six women vomited even during the rescheduled OGTT. Most participants followed the advice and were prepared for drawing two samples (fasting and post challenge). Only two respondents left the study centre after drawing the fasting sample, as they did not want to complete the post-pyramidal test.

Feasibility of Follow-up visits

The number of follow-up’s completed soon after delivery was 347(40%).Subsequent follow-up’s coincided with the immunization visits were 79, 54, 82, 45 and 43at their 6th/10th/14th weeks, 9th month and one year respectively. Delivering at other hospitals and at their native place (199) are the foremost reasons for loss to follow ups. In addition, two women delivered at home, two delivered preterm and 22 women refused for follow-ups. Five women delivered babies with stillbirth and one woman died of heart attack within3 hours of delivering the infant. Two women have children with developmental challenges and the parents were not ready to divulge any information during follow up 

Discussion

Our study which was done on a pilot basis demonstrates the feasibility of recruiting a reasonably large number of pregnant women from the public health facilities of Bengaluru , With over 80% of women enrolled after initial screening and 90% of enrolled women completing the OGTT, it is feasible to set-up birth cohort in public health facilities in Bengaluru. However, the drop-out rates during the follow up visits were high, showing that it is not feasible for concurrent placement of followups with the immunization visits. With over an hour required for the questionnaire administration and data entry, this study also suggests that the length of the paper-based questionnaire has to be reduced. Following up a larger number of women in public health facilities might require innovative methods. The recruitment and testing are completely feasible for initiating birth cohort in public health facilities. Barring the women who did not meet the eligibility criteria, only 2.5% were not interested in participating in the study. Of the enrolled women, 63 (7%) did not complete the questionnaire and anthropometry due to lack of interest and because they felt that questionnaire was too exhaustive.

Given that most public health facilities use only random blood sugar test, the feasibility of conducting OGTT in public health facilities is encouraging. Qiao et al examined random capillary blood glucose (RCBG) as a screening test for T2DM and followed it up with OGTT in the whole population. They reported a sensitivity of only 40% in women while specificity was 84% for a value of 5.8 mmol/L.15 The sensitivity of venous fasting plasma varies between 40% and 65% while the specificity is greater than 90% for Fasting Plasma Glucose (FPG) values ranging from 6.1 – 7.8 mmol/ L.^16,17 Petersen et al. found FPG variance from day to day to be 12–15%.¹⁸ The WHO recommends OGTT for diagnosing gestational hyperglycemia.¹⁹ In a prospective multicenter study with 5235 pregnant women, diagnostic testing with a 2-hour 75-g oral glucose tolerance test, a 2.4% prevalence of gestational diabetes mellitus was observed. Sensitivity and specificity of the model was 80.6 (73.7-87.6) and 64.8 (63.5-66.1), respectively (95% CIs).²⁰ The HAPO study (Hyperglycemia and Adverse Pregnancy Outcome) was successful in using a 2-hour 75g OGTT in 25,505 pregnant women and recommended universal screening of pregnant women with this test.²¹ Since, 88% women underwent the test in our study, it demonstrated that OGTT could be implemented as routine test in public health facilities. The minor proportion of women (32 respondents) having adverse events like vomiting, nausea, fainting or tiredness concurs with available evidence.^3,22-26 Though most pregnant women do not have side effects from the OGTT per se, some feel nauseated, sweaty or light-headed during the test, especially during the drawing of blood samples or before or after they drink the glucose solution. This poses a challenge to conduct the test. However, for various studies involving these tests, there is no sufficient evidence on whether or not the OGTT are triggering these side effects in pregnant women. Earlier, we had shown that majority (92%) of the doctors in primary health care settings have poor knowledge about the cut off values of glucose tests.³ Hence, the feasibility study suggests that the public health facilities can offer OGTT, thereby, the doctors can be enabled with confirmed diagnosis.²⁷

Preventing the lysis of glucose and maintaining the blood samples in cold chain during transport is important for the valid study results. The study showed the feasibility of centrifuging and refrigerating all the samples sent to the central laboratory for analysis within 1 hour. This aspect will remain crucial for expansion of any field-based study. Guidelines for plasma glucose analysis in accredited laboratory suggests blood plasma should be separated from the cells within 60 minutes, if it cannot be separated a tube containing a glycolytic inhibitor such as sodium fluoride should be used for collecting the sample.²⁸ Another recommendation suggests in order to minimize glycolysis, sample tube should be immediately placed in ice–water slurry, and the plasma should be separated from the cells within 30 min, if this is not possible a tube containing a rapidly effective glycolysis inhibitor (citrate buffer) should be used for collecting the sample. Tubes containing only enolase inhibitors (sodium fluoride) is not reliable to prevent glycolysis.²⁹

For any significant expansion of the study, it is important that food frequency questionnaire is removed for better utilization of limited time available. Studies reveal that food-frequency questionnaire do not provide reliable estimates of absolute amounts of dietary fats or cholesterol.³⁰ The degree of measurement error associated with nutrient estimates calculated from food frequency questionnaires leads to important underestimates of relative risks, and errors in other confounding nutrients. These errors will also result in inappropriately narrow confidence intervals. As per Feunekes GI et al, use of food frequency questionnaire and dietary history is not inaccurate method of assessing amount of dietary intake. These methods depend on the memory of the respondents and their ability to provide correct estimate of intake.³¹ In addition, as the respondent is likely to stay with her parents during pregnancy, the food frequency may not serve as a proxy for the sustained nutrition in the pregnant women. Generally, the food habits would vary from her husband and parent's place. Since FFQ is administered in 2nd trimester, the reference for the reporting data is first trimester, which is characterized by morning sickness, food intolerance and allergy. Hence FFQ will not serve as a reliable tool to estimate routine food habits. Instead, exploring the dietary and recording 24-hour diet recall could be beneficial. Data pertaining to contextual information such as type of ingredient used, cooking method and frequency of eating would be more beneficial.

Our study indicates that following up larger number of women and their children in public health facilities is very difficult. There are several reasons for the difficulty. Women are available in the hospital for only3 days in case of normal delivery and 7 days in case of caesarean sections. Although it was planned for timely follow-up, some women were sent home on the same day of delivery leading to imminent loss to follow-ups. Also, many respondents were routed to other hospitals for delivery as they were carrying out the renovation works in the operation theatre and labor ward. Also, the study sites were secondary level health care institutions and used to refer pregnant women with complications to tertiary level, thereby increasing the propensity of loss to follow-up at the time of delivery. Followups during immunization visits are very low, as only those respondents who stay very close to the vicinity turn up at the study sites for immunization of the infant. Although women choose the hospitals (study centers) for the delivery, they find it easier to visit the primary healthcare centre further close to their home for immunization. A minor proportion of them do not visit any health facility, as they do not understand the importance of immunization. A study conducted in North India reveals that poor education status of the parents is a main reason for lower immunization coverage.³² Theplan for scheduling 4 follows ups in the first year is not feasible and will lead to many loss to follow ups as respondents tend to keep moving from their home to their maternal home very frequently as they get lot of support in taking care of their infants. Moreover, we observed that the skin fold thickness of infants did not differ much in the 6th, 10th and 14th week, the average skin fold thickness in 6th, 10th and 14th week were 7.3mm. 8.1 mm and 8.4 mm respectively. Hence, one follow-up visit can be planned within the first year.

The result from the feasibility study has provided several pointers for planning successful followup of pregnant women and children. First, we could trace some of our respondents with the support of link workers to trace them. Second, the research group has visited anganwadis (early childcare centers) and health camps conducted by government in research localities to the track the study respondents. Third, several innovative methods need to be tried for ensuring regular follow-up. These include building the rapport of research team with pregnant women. Frequent and warm interactions, enquiring about their wellbeing, assisting them with their hospital registration or in obtaining the government amenities, wishing children on their birthdays, providing the infants photo album from current and previous follow up visits are some of the methods which can be helpful in maintaining good follow up rates. The pilot study has provided important action points for reducing the loss to follow-ups. India has implemented Mother and child tracking system (MCTS) for tracking maternal and child health beneficiaries in India’s public health system, and improving service delivery planning and outcomes³³.The health workers, in turn, ensure that pregnant women receive adequate ante-natal and post-natal care besides encouraging institutional deliveries. It would be beneficial for any studies to work with health authorities in tracking the study respondents and in ensuring the regular visits. In the instances, where the study respondents find it difficult to visit the hospital for follow ups, it would be better to involve the self-help groups or anganwadis (early childcare centers) to track and complete the visits. In addition, greater awareness is needed in order to encourage women to get tested and visit the hospital for ensuring health services to themselves and their offspring. The experience of this study indicates that pregnant women attending public health facilities in Bengaluru (from different communities), can be recruited and tested. For follow ups it is necessary to establish strong network with the pregnant women and health workers.

Conclusion

Our study found that it is feasible to recruit, administer OGTT to large number of pregnant women in setting up a birth cohort in public health facilities. However, there are several challenges in ensuring regular follow-up. This work should inform and assist researchers undertaking or planning similar studies in public health facilities. As GH is a ‘silent’ disease, the universal screening of pregnant women with OGTT should be carried out.^34,35 It will be useful to align the cohort studies with tracking mechanisms for ensuring regular follow-up.³⁵ It is important to have the support of health workers and necessary logistics for an efficient cohort study. Though international agreement on the clinical criteria and method of detection of GH has been elusive, our study has provided the utility of conducting it in public health facilities.^34-36 With careful planning for ensuring regular followup, the expansion of the cohort study can provide evidence to formulate guidelines by clinicians and policymakers to manage differing levels of blood glucose in pregnant women.

Funding:“This work was supported by a Well come Trust Capacity Strengthening Strategic Award to the Public Health Foundation of India and a consortium of UK universities”.

Contributor-ship statement: GRB was involved in the conception and design, drafting the article. GVSM was involved revising it critically for important intellectual content and final approval of the version to be published. VGM was involved in the contribution to the field sites and final approval of the version to be published. AVK was involved in revising it critically SK was involved in conception and design, revising it critically for important intellectual content and final approval of the version to be published.