Trends in TSH, free T4, and anti-thyroid peroxidase and treatment status: Canadian Health Measures Survey 2012 to 2015

Abnormality identification for blood measures

We assessed TSH, free T4, and anti-thyroid peroxidase for abnormality, values higher or lower than upper or lower reference limits, respectively. The reference ranges for TSH were 0.4 to 4.8 mIU/l for adults and 1.7 to 9.1 mIU/l for those aged 21 weeks to 20 years.[15] The reference ranges for free T4 were 13 to 27 pmol/l.[15] The upper limit of the anti-thyroid peroxidase reference range was 35 U/ml.[16]

Treatment status identification

We defined the treatment status according to the use of medication and the chronic condition diagnoses reported by participants.[17-20] We estimated the treatment status in four categories: treated, probably treated, potentially treated, and untreated (see Table 1). To avoid over-identification, we defined untreated individuals as those who had not reported the use of any medications or any related chronic conditions defined by the CHMS (any chronic conditions, cardiovascular disease, high blood pressure, abnormal lipid profile, stroke, kidney disease, liver disease, thyroid disease, diabetes, cancer, eating disorder, or chronic obstructive pulmonary disease). We regarded individuals as treated only if they were diagnosed with thyroid disease or who were taking thyroid prescription drugs.

Individuals were probably treated if they were diagnosed with the above-mentioned chronic conditions that might affect clinical measures or were taking any prescription drugs. Individuals were regarded as potentially treated if they reported any chronic conditions or were taking any prescription or over-the-counter drugs.

We reported medication use based on the Anatomical Therapeutic Chemical (ATC) classification system.[9, 21] Medication in the CHMS data was labelled as prescription, over-the-counter drugs, and health products.[9] Over-the-counter drugs and health products were regarded as non-prescription.[9] We extracted the ATC codes that represented prescription or over-the-counter drugs from the medication file of each cycle. In Table 1, the first and second levels of the ATC codes that represented thyroid medications were indicated as “G03.”

Data analysis

We conducted a trend analysis using all the variables available from the CHMS.[8, 9, 22] First, we summarized these variables using basic characteristics, such as maximal and minimal values, and we linked the survey and bootstrap weights to them.[22] Then, we identified the missing values and the values beyond the detection levels for all the variables.[10] We recorded basic statistics, such as the weighted mean, and the 25th and 75th percentiles.[10] We introduced clinical reference ranges to identify Canadians with abnormal levels of thyroid measures. We labelled four treatment statuses based on chronic conditions and medication use. We used the CHMS cycles as the proxy measure for time points in our trend analysis. Using ratios, we compared the distributions of TSH, free T4, and anti-thyroid peroxidase in CHMS cycle 2 (2012 to 2013) to their distributions in CHMS cycle 3 (2014 to 2015).[9, 10, 23] We described the trends with ratios with 95% confidence intervals (CIs). The 95% CIs of the ratios in the subsequent cycles covering 1 indicated insignificant changes in the levels of thyroid measures after cycle 3. We conducted the data analysis with R (v3.20)[24] and RStudio (v0.98.113).[25]

Proportions of abnormality in blood measures

The proportions of Canadians with levels higher or lower than the reference ranges differed for TSH, free T4, and anti-thyroid peroxidase. There was an increase in the proportions of Canadians with high levels of anti-thyroid peroxidase and low levels of TSH. There was a decrease in the proportions of Canadians with high levels of free T4 and TSH, and low levels of free T4. The proportions of Canadians with clinically high levels of free T4, anti-thyroid peroxidase, and TSH were 0.005 to 0.005, 0.30 to 0.43, and 0.03 to 0.017 for cycles 3 to 4, respectively. The proportions of Canadians with clinically low levels of free T4 and TSH were 0.18 to 0.11 and 0.02 to 0.021 for cycles 3 to 4, respectively. The lower reference limit for anti-thyroid peroxidase was not defined, and the proportion with clinically low levels was not available.

Treatment status for clinically high levels

The four estimated treatment status varied for TSH, free T4, and anti-thyroid peroxidase. Due to an insufficient number of unweighted sample sizes, the proportions of Canadians being treated with high levels of free T4 were not reported in Table 2. The proportions of Canadians who were untreated—based on drug use and self-reported conditions—were all less than 5% of Canadians with high levels of free T4, anti-thyroid peroxidase, and TSH: 0 to 0.009, 0.02 to 0.03, and 0.01 to 0.049 for cycle 3 and 4, respectively.

The proportions of Canadians treated with high levels of TSH and anti-thyroid peroxidase were 0.31 to 0.44 and 0.15 to 0.11 for cycles 3 to 4, respectively; whereas sufficient sample sizes were unavailable for estimating this proportion for free T4.

The present study uniquely identified two treatment statuses—probably treated or potentially treated. Among those with high levels of free T4, anti-thyroid peroxidase, and TSH, the proportions of being probably treated were high: 0.34 to 0.92, 0.90 to 0.90, and 0.87 to 0.86 for cycles 3 to 4, respectively. Among Canadians with high levels of free T4, anti-thyroid peroxidase, and TSH, the proportions of being probably treated were high: 1 to 0.99, 0.98 to 0.97, and 0.99 to 0.95 for cycles 3 to 4, respectively.

Treatment status for clinically low levels

The treatment status for clinically low levels of anti-thyroid peroxidase was not defined. The proportions of Canadians with low levels of free T4 and TSH were 0.18 to 0.11 and 0.020 to 0.021 for cycles 3 to 4, respectively. The proportions of untreated Canadians with low levels of free T4 and TSH were 0.025 to 0.007 and 0 to 0.01 for cycles 3 to 4, respectively. The proportions of Canadians being treated who had low levels of free T4 and TSH were 0.09 to 0.04 and 0.44 to 0.41 for cycles 3 to 4, respectively.

The proportions of Canadians being probably treated were high for free T4 and TSH. Thus, the proportions of Canadians with low levels of free T4 and TSH being probably treated were 0.89 to 0.93 and 0.87 to 0.91 for cycles 3 to 4, respectively. The proportions of Canadians being potentially treated were almost 1 for free T4 and TSH. Thus, the proportions of Canadians with low levels of free T4 and TSH being potentially treated were 0.97 to 0.99 and 1 to 0.99 for cycles 3 to 4, respectively.

Treatment and clinically abnormal levels

One feature of the present study is its four treatment statuses indicating the intensity of medical interventions that Canadians have received. Moreover, the proportions of the four treatment statuses that identified TSH, free T4, and anti-thyroid peroxidase were different. The first finding based on treatment statuses is that thyroid medications are rarely used by Canadians with high levels of free T4. An increase in the levels of free T4 has been linked to increased risks of developing any solid cancer, lung cancer, and breast cancer, regardless of the TSH levels.[30] This outcome may be due to the activation of ontogenetic pathways and tumor-growth-related genes.[30] The second finding is that the majority of Canadians with abnormal levels of TSH, free T4, or anti-thyroid peroxidase were probably treated (were taking prescriptions or diagnosed with chronic conditions), more than 86%, except free T4 in cycle 3 (34%). Whether this is due to an increase in the treatment coverage, changes to clinical practice, or differences in sampling methods needs to be confirmed with future CHMS cycles.

The third finding based on treatment statuses is that although individuals may be treated, the levels of thyroid measures remain abnormal, i.e., inadequately controlled. The proportions of individuals with inadequately controlled levels of lipid profile in the general public have been used to assess the performance of health care systems.[18] TSH, free T4, and total T3 are important measures to follow-up with patients who have a thyroid condition.[31, 32] Our estimates of the proportions of uncontrolled high or low levels of TSH are more than 31% or 40% among those with high or low levels of TSH, respectively. The proportions of uncontrolled high or low levels of free T4 are less than 1% or more than 4% among those with high or low levels of free T4, respectively. We did not find similar statistics reported by other researchers. Whether these present study statistics are useful for assessing the efficacy of thyroid condition intervention remains to be investigated.

The fourth finding based on treatment statuses is that the association of medical interventions with the proportions of clinically high or low levels of TSH, free T4, and anti-thyroid peroxidase varies. Various factors can modify treatment efficacy, such as intervention duration, treatment doses, frequency,[33], and compliance to medical interventions.[20, 34, 35] It remains unclear whether the proportions of inadequately controlled abnormality of thyroid measures is a useful indicator for assessing intervention adequacy for those patients with abnormal levels of TSH, free T4, or anti-thyroid peroxidase.

Last, opportunities may exist to intervene based on the association of treatment statuses and abnormality in blood measures. The differences in the proportions of being treated and potentially treated among Canadians with clinically high or low levels are more than 50% for TSH, free T4, and anti-thyroid peroxidase in cycles 3 and 4. The difference between being probably treated and potentially treated is taking over-the-counter medications, which suggests that most Canadians with abnormal levels of TSH, free T4, or anti-thyroid peroxidase were taking some medication. With respect to TSH, this suggests that more than 94% of Canadians with clinically high or low levels may be targeted for interventions through their regular access to medication. Several pharmacist-led initiatives have been designed to screen conditions for treatment, for example, sexually transmitted infections,[36] sleep apnea,[37] asthma, obesity, and hypertension.[38] Also, pediatric patients have been identified by pharmacist-led screening programs.[38] Our results may help to refine the directions and magnitudes that health authorities can use to design population screening programs when necessary.

Overdiagnosis of thyroid conditions?

Korean researchers used a 15-fold increase in thyroid cancer diagnosis prevalence and stable thyroid cancer mortality to demonstrate the possible extent of overdiagnosis of thyroid cancer.[6] The present study was not designed to directly assess an overdiagnosis or undertreatment of a thyroid condition, but the trends in, and the treatment statuses of, thyroid measures can be used as the foundation for assessing these issues. Thus, the present study findings can be informative for planning an investigation of the overdiagnosis or undertreatment of thyroid conditions. First, it may not be adequate to assess the trends in thyroid treatment solely based on the use of a thyroid prescription (ATC codes beginning with G03) because the use rate is low among those with abnormal levels of TSH. Second, the levels of TSH, free T4, and anti-thyroid peroxidase remain similar throughout the CHMS cycles 3 and 4. These trends provide important baseline data for future monitoring. Whether the trends in TSH, free T4, and anti-thyroid peroxidase may be changed or altered by other factors can be studied with such data.

Last, the assessment of overdiagnosis is difficult with information based only on blood measures and treatment status at two time points. A long-term follow-up and major health outcomes, such as mortality, will be needed for assessing an overdiagnosis.[6] At the same time, the measures of the CHMS are subject to changes in medical practices and sampling frames, or to measurement errors.[8] Without obvious signs in CHMS cycles 3 and 4, this issue needs to be monitored in the future.

Sick population versus sick individuals

The results of the present study show that large proportions of Canadians with clinically high or low levels of anti-thyroid peroxidase and free T4 may highlight the shortcomings of current health care that focus more on sick individuals than the signs of increasing risks to thyroid health in the general population.[39] Previous Canadian studies have reported on only the proportions of patients with clinically high or low levels of TSH.[27] With respect to the Canadian population, the distributions of TSH, free T4, and anti-thyroid peroxidase have not been well studied, let alone the trends of their distributions. Our research group has been developing tools to analyze trends in national surveys[8-10] and to demonstrate the importance of trends in blood measures and environmental chemicals.[8-10] If the increasing risks of thyroid conditions are limited only to certain subgroups, preventive strategies should monitor and focus on individuals with such characteristics, i.e., the high-risk strategies described by Geoffrey Rose.[39] However, the increase in risks applies to the general population. This strategy will require policies that do not focus on certain individuals, i.e., the population strategies described by Geoffrey Rose.[39] We think the results of this trend analysis provide a good foundation for a discussion on how to reduce the risk for Canadians to develop long-term consequences of untreated or undertreated thyroid conditions.

Limitations

Fluctuations of large magnitudes in other blood measures or environmental chemicals have occurred throughout CHMS cycles 1 to 4.[8] Besides the changes in the distributions of blood measures across time, other factors may lead to different levels of blood measures or various proportions of treatment statuses over time, such as differences in sampling frames, quantification methods, or precisions; and changes in detection levels.[8] Moreover, a lack of overall significant trends in population levels may not mean a lack of significant changes in blood measure levels among subpopulations. The CHMS data are not designed to track or follow-up with individuals with specific characteristics or smaller geographic areas[11-13], and so it may not be adequate to conclude that significant trends in all subpopulations are lacking.