Study design and Setting
This was a cross-sectional study to determine the prevalence and factors associated with hypothyroidism among children with sickle cell anemia aged 6 months to 17 years attending the sickle cell clinic in Mulago hospital between September and October 2020. Mulago Hospital is Uganda’s National Referral and Teaching Hospital for Makerere University. It receives patients referred from health facilities within and outside Kampala District. The SCC runs 5 days a week, has more than 15,000 registered patients, receives about 60-80 patients daily and over 70 new patients monthly. The majority (75%) of the patients are below 18 years and most are from Kampala and the surrounding districts. A third of those seen daily come in with acute illness or crises while the rest come for routine check-up and drug refills. There is no routine screening for thyroid dysfunction at the clinic.
We included children with confirmed sickle cell anemia aged 6 months to 17 years who attended the sickle cell clinic during the study period, provided assent for those 8 years and above and whose caregivers provided informed consent. We excluded children on treatment for hypothyroidism with levothyroxine or on anti-thyroid drugs such as carbimazole and propylthiouracil and those who were too ill to withstand study procedures.
Sample size calculation
We used the formula by Scheaffer, Mendenhall III (15) to obtain the minimum sample size required to determine the prevalence of hypothyroidism in children with sickle cell anemia.
n = sample size
deff = Design effect=1
N = population size (N= 40 patients per day x 20 working days x 3 month study period x 75% below 18 years= 1800 )
pˆ = anticipated prevalence of hypothyroidism in children with sickle cell anemia= 50% giving the largest sample size, since there were no studies done in a similar population.
qˆ = 1 - pˆ = 0.5
d = desired absolute precision or absolute level of precision =5%, the study had 80% power
Assuming 5% non-response rate adjusted sample size = 332
Consecutive sampling was used; that is: every child with sickle cell anemia aged 6 months-17 years who attended the SCC during the study period and met the inclusion criteria was enrolled until the sample size was achieved.
A pre-tested structured questionnaire was used to obtain the socio-demographic data, nutritional and past medical history. A detailed physical examination for signs of hypothyroidism was performed by the principal investigator or research assistant. Anthropometric measurements which included weight, height/length and mid-upper arm circumference were also recorded.
A venous blood sample of 3ml was drawn from either the ante-cubital or femoral vein using a needle and syringe and put into a plain red top vacutainer tube. This was used to measure serum thyroid stimulating hormone and free thyroxine. Fasting before the blood collection was not required. The blood samples were transported to a quality laboratory that is MBN laboratory in Kampala, within 1 hour of collection. After cross-checking using the generated sample log, the samples were then centrifuged and aliquoted and analyzed on a daily basis. TSH and freeT4 levels were measured by electrochemiluminescent immunoassay technique using a fully automated COBAS 6000 ROCHE HITACHI machine from Germany. The definitions of hypothyroidism were;
TSH level >9 mIU/L and free T4 <0.6 ng/dL or Normal TSH and free T4 < 0.6 ng/dL were considered as clinical hypothyroidism. TSH ranging between 4.5 and10 mIU/L with normal age appropriate T4 levels that is: 6-11 months [0.9-2.0ng/dl]; 1-5years [1.0-1.8ng/dl];6-10 years [1.0-1.7ng/dl] and 11-17 years [1.0-1.6 ng/dl indicated sub-clinical hypothyroidism.
All children found with hypothyroidism were referred to the Paediatric endocrine clinic for further management.
A complete blood count (run by a SYSMEX XNL-450 machine) was also obtained from the clinic as part of their routine care and clinic protocol. A recent hemoglobin electrophoresis (within 6 months before enrollment) was also used if available to record the fetal hemoglobin levels.
The questionnaires were numbered with serial numbers that corresponded to the labels on the blood samples. Questionnaires were checked for accuracy and completeness, and stored in a safe place under lock and key. Data was entered into the computer that is password protected using double entry method and cleaned using Epidata version 3.1 and exported to Stata version 14 for analysis.
The study had 80% power, with an absolute error between the estimated and true value of 5% and with 95% confidence intervals.
Baseline characteristics were summarized as mean, standard deviation, median and interquartile range for continuous variables and frequency and proportions for categorical. The proportion of children with hypothyroidism was obtained by dividing the number of children with hypothyroidism by the number of all children with sickle cell anemia aged 6 months-17 years who were enrolled in the study. Simple logistic regression was used to test the association between hypothyroidism and independent variables individually at 5% level of significance and any variable that achieved a p< 0.2 was considered for multivariable analysis. Multiple logistic regression models were used to assess the simultaneous association between the dependent variable and independent variables. Backward elimination using an inclusion criteria p < 0.2 was used to build model of best fit. Goodness of fit test was conducted at 5 % level of significance.
Interaction and confounding were assessed.