I I SPECIALARTICLE Admission Screening by Thyroid Function Tests in an Acute General Care Teaching Hospital LESLIE J. DEGROOT, M.D., GILBERT MAYOR, M.D.,
Chicago,///inois
To determine the incidence of unrecognized thyroid disease among admimaions to a large acute care university teachinr hospital, 364 sampies taken on consecutive admissions were assayed for thyroid-stlmulating hormone (TSH) and free thy ~ - e index (l~I). Patients with abnormnl test results were further evaluated by determin~tion of antlmlcresomal and antithyroglob-lin antibodies, and c h R m were reviewed for evidence of prior diagnosis of thyroid disease, especially severe illness, drug treatment that might affect thyroid function tests, and prior diagnosis of thyroid diseas~ Results of subsequent thyroid function tests performed during the patient's hospitali~tion were correlated with the admlmlon serum assays, and data on subsequent testing during the following 6 months were also obtained. A total of 3.9% of patients had significantly depressed TSH, and 11.1% of values were significantly elevated. A total of 11~1% of patients had significantly low FTI values, and 1% had significantly elevated values, A total of 7.4% appeared to have the euthyroid sick syndrome, 5~% appeared to have unrecognized or undertreated primRmy thyroid failure, 6% had apparent subclinical hypothyroidism, 2% were thyrotoxic, and 2~% (all women) had suppressed TSH levels for inapparent reasons. Limltin~ testing to patients over 49 years of age, or to women, would have mimed many individnAIA with abnormal test results. Considering widespread availability of tests, relative costs, and value of the information obtained, it is suggested thet the FTI determ;n=tion would provide an appropriate screening test for patients in a population such as this entering a large, acute care general hospital.
From the Thyroid Study Unit, Department of Medicine, The University of Chicago, Chicago, Illinois. This work was supported by United States Public Health Service Grants DK13377 and DK27384, the March of Dimes Birth Defects Foundation, Boots Pharmaceuticals Inc., and the David Wiener Research Fund. Requests for reprints should be addressedto Leslie J. DeGroot, M.D., Thyroid Study Unit/Mail Code 3090, The University of Chicago, 5841 South Maryland Avenue, Chicago, Illinois 60637. Manuscript submitted September 26, 1991, and accepted in revised form August 19, 1992.
558
he value of screening populations for undiagnosed thyroid disease has been investigated in T a variety of settings including elder-care institutions, Veterans Administration hospitals, acute general hospitals, and outpatient clinics [1-5]. Typically, 2% to 10% of subjects are found to have undiagnosed thyroid disease. While the medical benefits of detecting even mild thyroid disease are presumed to be self-evident, the analysis of economic benefit has been difficult but appears to favor screening [6]. We note that thyroid function tests are believed to already account for nearly a billion of the several hundred billion dollars spent on health care each year in the United States [6]. Nevertheless, some level of laboratory screening of new admissions to acute care hospitals is commonly performed. We reasoned that substitution of a relatively inexpensive thyroid function test in an admission "chemistry profile" might be cost-effective, and perhaps could replace some other component of the profde. For example, the serum phosphate, a typical component of the screening chemistry profile, rarely provides significant information, although it obviously may be of extreme importance in renal disease, ketoacidosis, and other disorders in which attention is specifically directed to serum phosphate determination. To explore this question further, we performed thyroid function tests on nearly 400 consecutive, unselected admissions to an acute care general teaching hospital, analyzed the relationship of the abnormalities discovered to characteristics of the population, and compared the utility of different assays for screening.
METHODS With the approval of the Institutional Review Board, discarded serum samples were obtained from the hospital chemistry laboratory. The first serum samples sent to the chemistry laboratory, within 24 hours of admission of a new patient, were collected until 400 consecutive samples had been recovered. For technical reasons (duplication, insufficient serum, contamination, inadequate identification, and so forth), some were excluded, leaving 364 samples available for analysis. On this group of sera, thyroid-stimulating hormone (TSH) was assayed using an immunoradioactive assay (Serono,
November 1992 The American Journal of Medicine Volume 93
SCREENING BY THYROID FUNCTION TESTS / DeGROOT AND MAYOR
TABLE I
TABLE II
Distribution of Total Population by TSH Values
Distribution of Total Population by FTI Values
TSH (I~U/mL)
Frequency
%
Age (y)* < 50 > 49
0-0.2 0.3 0.4-4 4.1-5 5.1-8 >8
14 9 280 17 24 20
3.9 2.5 76.8 4.7 6.6 5.5
5 7 130 7 7 3
Total
364
100.0
159
9 2 150 10 17 17 205
FTI (units)
Sex* M
F
0 3 136 4 7 5
14 6 144 13 17 15
155
209
TSH= thyroid-stimulatinghormone, *Distributionof resultsnonrandomin relationtoageandsex(p < 0,05).
49 16
25
Sex(NS) M F 20
21
5-5.9
52
14.3
27
25
23
29
6-10.5
258
70.8
110
148
108
150
10.6-11.5
9
2.5
4
5
3
6
>11.5
4
1.1
2
2
1
3
364
100.0
159
205
155
209
Total
FTI= freethyroxineindex;NS= distributionofpatientsbyageandsexis randomin relationto FTI,
Allentown, PA), thyroxine (T4) was assayed using an immunoassay (Diagnostic Products, Los Angeles, CA), and free thyroxine index (FTI) was derived as reported using a resin T4 uptake method [7]. In normal subjects studied in our laboratory, serum TSH levels display a bell-shaped distribution, with 98% of values ranging from 0.4 to 4 pU/mL, which was taken to be the normal range. Serum TSH levels were separated into six categories: 0 to 0.2 #U/mI~assumed to be significantly low; 0.3 pU/mL--considered borderline; 0.4 to 4 pU/mL--considered normal; 4.1 to 5 pU/mL--considered minimally elevated; 5.1 to 8 ~U/mL--considered elevated; and greater than 8 pU/mL--considered seriously elevated. T4 values were similarly categorized. Values less than 4 pg/dL were considered significantly low, 4.4 to 4.9 ~g/dL borderline, 5 to 12 pg/dL normal, 12.1 to 13 pg/dL minimally elevated, and greater than 13 ~g/dL significantly elevated. Serum FTI values were broken into corresponding categories: less than 5, indicating significant depression; 5 to 5.9, borderline; 6 to 10.5, normal; 10.6 to 11.5, borderline elevated; and greater than 11.5, significantly elevated. Age and sex of patients were recorded. Data from patients with significantly low or elevated FTI or TSH were further reviewed by analysis of laboratory records and hospital charts. The hospital course was analyzed to determine the presence of extremely severe illness on admission, use of medications that might depress or elevate the TSH or FTI (dopamine, prednisone, diphenylhydantoin, propylthiouracil, thyroxine), and previously or concurrently diagnosed thyroid disease. In this subgroup of patients, serum was also analyzed for the presence of antimicrosomal and antithyroglobulin antibodies by hemagglutination assay [8]. Results were tabulated in a "Dbase" (Ashton-Tate, Culver City, CA) program and further analyzed by program "Stata" (Computing Resource Center, Los Angeles,
CA). Group differences observed in linear regression or ×2 analysis were considered significant if the p value was 1/20. *Distributionof resultsrandomin relationto totalpatientgroup. tOistributionof resultsnonrandomin relationto totalpatientgroup.
TABLE IV Characteristics of Patients With Abnormal FTI or TSH, Grouped by FTI Values Age* FTI
Frequency
Sext
ESS*
%
> 49
< 50
M
F
+
-
0
TDX* 1
2
0
Drug* 1
2
Antibodies* +
11.5
41 7 32 3 4
47.1 8.1 36.7 3.5 4.6
25 6 24 1 2
16 1 8 2 2
20 5 3 1 1
21 2 29 2 3
23 4 13 0 1
18 3 19 3 3
37 6 25 2 1
4 1 6 1 2
0 0 1 0 1
33 6 21 2 2
5 1 6 0 1
3 0 5 1 1
30 3 18 3 1
9 3 10 0 1
Total
87
i00.0
58
29
30
57
41
46
71
14
2
64
13
10
55
23
FTI = freethyroxineindex;TSH = thyroid-stimulatinghormone;ESS= euthyroidsicksyndrome(verysevereillnessasestimatedby reviewof clinicalrecord);TDX= 1--prior diagnosisof Hashimoto's thyroiditisor hypothyroidism,or 2--prior diagnosisof hyperthyroidism;Drug= 1--prednisone,diphenylhydantoin,aspirin,dopamine,propylthiouraciladministration,or 2--thyroxineadministration; Antibodies= microsomalorthyroglobuiinhemagglutinationtiter > 1/20. *Distributionof resultsrandomin relationto totalpatientgroup. t Distributionof resultsnonrandomin relationtototalpatientgroup.
TABLE V Distribution of Total Population by TSH and FTI Values TSH (I~U/mL) 0-4.9
5-5.9
FTI (units) 6-10.5 ]0.6-11.5
>11.5
Total
0-0.2 0.3 0.4-4 4.1-5 5.1-8 >8
1' 1" 25* 111 211 1 llr
0 0§ 47§ 2§ 511 211
101 7§ 201§ 14§ 164 64
1 0§ 6§ 0§ 1 1
2' 1 1 0 0 0
14 9 280 17 24 20
Total
41
56
254
9
4
364
TSH= thyroid-stimulatinghormone;FTI = freethyroxineindex. *Euthyroidsicksyndrome(7.4%of patients). tSuppreesedTSH(2,8%of patients). tToxic(1,1%of patients). §Paranormal. IIHypothyroid(5.8%of patients). ISubclinicalhypothyroidism(6.0%of patients).
Characteristics of Patients With Either Abnormal TSH or FTI To further characterize the group of patients with abnormal TSH or FTI values, hospital records were evaluated for evidence of extremely severe ill560
ness, prior thyroid diagnosis, use of certain medications, or presence of autoimmune thyroid disease. As shown in Table III, patients with low TSH uniquely were female. Nearly half (6 of 14) had a previous diagnosis compatible with autoimrnune thyroid disease, and 4 of 15 were taking medications that might suppress TSH. Not surprisingly, antithyroid antibodies were detected in 10 of 19 patients with TSH elevated above 8 #U/mL. Analysis of age, sex, severity of illness, prior history of thyroid disease, medications, and antithyroid antibodies, in relation to serum FTI, indicated a striking excess of males among patients with a reduced FTI (Table IV). Categorization of the Population by TSH and FTI Values In theory, determination of TSH and FTI should correctly place most patients into diagnostic categories. This distribution is shown in Table V. Two hundred seventy-seven of the patients fell into a normal or "paranormal" category. Twenty-seven patients had a low FTI with TSH normal or sup-
November 1992 The American Journal of Medicine Volume 93
SCREENING BY THYROID FUNCTION TESTS / DeGROOT AND MAYOR TABLE VI Relation of Clinical Factors to Categorizationof Patients by FTI and TSH Values ApparentEuthyroid
ApparentPrimary
SickSyndrome
Hypothyroidism
Suppressed
TSH
Subclinical
Apparent
Total Group
Hypothyroidism Hyperthyroidism
Priorthyroiddiagnosis None Hashimoto'sthyroiditis Graves'
26 1 0
17 4 0
6 4 0
16 6 0
0 2 0
65 17 0
Severeillness + -
14 13
13 8
4 6
9 13
0 2
40 42
Medication None Category1 (TableIII) Category2 (TableIII)
22 4 1
17 2 2
4 3 3
17 3 2
1 1 0
61 13 8
Antibodies + -
3 23
9 10
4 5
6 13
1 0
23 51
Age 49
12 15
5 16
5 5
3 19
0 2
25 57
Sex M F
16 11
9 12
0 10
3 19
0 2
28 54
FTI = free thyroxine index; TSH = thyroid-stimulating hormone.
pressed, suggesting the "euthyroid sick syndrome" (ESS) or, much less likely, hypopituitarism. This group comprised 7.4% of the entire population. As shown in Table VI, there was an excess of males in the ESS group, and 14 of 27 had unusually severe illness, but the patients were not significantly older than the entire group, and only 4 were taking drugs that might significantly alter FTI. Twenty-one (5.8%) had low FTI and elevated TSH, indicating primary hypothyroidism (Table V). Four of these patients had previously been recognized to have thyroid disease, but had not been given replacement therapy, or had been given it in inadequate amounts. These patients were older than the average, more (62%) had severe illness, and a higher proportion (47%) had positive antibodies, fitting with a probable diagnosis of Hashimoto's thyroiditis (Table VI). A total of 2.8% of patients, all women, had a norreal FTI with suppressed TSH. In three, medication might have been responsible, but in the remainder the cause is obscure. Perhaps it is a form of ESS. Six percent of patients had an elevated TSH with normal FTI, suggesting subclinical hypothyroidism; these patients were primarily older women, and 6 of 16 had some known prior clinical or laboratory finding suggesting they had Hashlmoto's thyroiditis. There was no excess of positive antibody assays in this group. Two patients had an elevated FTI and suppressed TSH, suggesting hyperthyroidism. One patient had been given excess
TABLE VII
Distributionof Total Patients by SerumT4 Values T4
(iLg/dL)
Frequency
%
>4 4-4.9 5-12 12.1-13 > 13
31 28 286 9 10
8.5 7.7 78.5 2.5 2.8
Total
364
100.0
T4 = thyroxine.
thyroid hormone for replacement of primary thyroid failure (Tables V and VI). Five patients with abnormal testsdid not fitinto any obvious diagnostic pattern.
Analysis of Serum T4 Values Analysis of serum T4 values showed that 8.5% of patients had a value less than 4 pg/dL, and 2.8% had a value greater than 13 #g/dL. This distribution is very similar to that of the FTI values (Table VII). Patient categorization by TSH and serum T4 values was evaluated (data not shown). While the distribution of patients in terms of numbers in each category was quite similar to that observed for FTI, on analysis it was found that the patients in the categories differed. For example, 11 of the patients classified as having ESS, on the basis of FTI and TSH,
November 1992 The American Journal of Medicine Volume 93
561
SCREENING BY THYROID FUNCTION TESTS / DeGROOT AND MAYOR
total population (including 45% of those with abnormal test results), another thyroid function test was ordered within 6 months. Obviously, other tests were probably performed in other laboratories as well.
TABLE VIII Clinical Review of Patients With Abnormal TSH TSH 8 TSH = thryoid-stimulatinghormone;T,L= thyroxine;PTU = propylthiouracil;FTI = fleethyroxine index;Rx= therapy,
were not present in the similar group classified on the basis of s e r u m T4. Five of the patients classified as having primary thyroid failure by FTI in Table V did not appear in this category when analyzed on the basis of serum TSH and serum T4. FTI is well known to be a better index of available thyroid hormone than serum T4. Because of this background and the apparent errors in categorization on the basis of s e r u m T4, further analysis of patients on the basis of s e r u m T4 values was not performed. Diagnosis and Follow-Up All seriously abnormal results were reported directly to the patient's physician, and reports were posted in the patients' charts. By analysis of laboratory records, it was found that in only 1% of the entire population (3% of the patients with abnormal tests) had the attending physicians ordered thyroid function tests prior to availability of the results from this survey. Thirty-seven patients had a second FTI determination following report Of the first sample. The correlation coefficient of the two assays was 0.74. In 11% the new FTI value removed the patient from the original diagnostic category. Thirty-nine patients (including 44% of those with an abnormal test result) had a second in-hospital TSH determination. The correlation coefficient was lower, 0.54, often because the second TSH leveis were much more elevated. In 13%, the .result shifted the patient to a different diagnostic category. Considering that some patients received specific treatment or had altered clinical conditions or treatments between tests, it is apparent that repetition verified the original observation in nearly 90% of the test cases. We also found that in 11.3% of the 562
November1992 The American J o u m ~ of Medicine
On the basis of assay of serum TSI-I and FTI, 6% of patients were found to have subclinical hypothyroidism, 5.8% to have probable primary hypothyroidism, and 3% to have an FTI less than 5 and TSH greater than 8, indicating an urgent need for thyroid hormone replacement therapy (Table I, Table V, and Table VIII). Overall, 20 of the 364 patients, or 5.5%, needed thyroid hormone replacement therapy, as judged by usual standards of medical management. Of the 13 patients with a significantly suppressed TSH, the cause was due to overreplacement with thyroid hormone in 3 and probably the ESS in 1, but the cause was inapparent in the majority of cases who were older women. Serum TSH assay identified as abnormal individuals with apparent primary hypothyroidism, mainly individuals without a known prior diagnosis, and a significant number of patients (20 patients, 5.5%) who might benefit from treatment. While this population had an increased proportion of individuals who were over age 50, and female, limiting testing to women above age 50 would have missed others: males (5 of 20) and younger patients (3 of 20) with significant hypothyroidism. Three of 15 patients with suppressed TSH appeared to have this finding on the basis of administered thyroxine. Screening by TSH assay also led to identification as abnormal a significant number (6%) of patients with subclinical hypothyroidism, and 2.8% with only suppressed TSH. These patients presumably do not need therapy at this time for a thyroid problem. Screening on the basis of FTI indicated that 11.3% had significantly low values and 1.1% had elevated values. The low FTI values were Rimost equally divided between patients who appeared to have ESS and patients with apparent primary hypothyroidism. FTI did not, by definition, identify patients with "subclinical hypothyroidism." Patients with abnormal FTIs were not clustered on the basis of age or sex. Drug effects did not explain a significant proportion of the abnormal results, and, somewhat surprisingly, there was no excess of extremely ill patients among the individuals with marked depression of FTI. Clearly, neither TSH nor FTI screening alone properly identified all patients with a need for therapy. For example, while TSH identified the individuals with apparent primary hypothyroidism, it also identified a group of individuals, equal in number,
Volume 93
SCREENING BY THYROID FUNCTION TESTS / DeGROOT AND MAYOR
TABLE IX Analysisof TSH/FTI Data If FTI Is Correct
If TSH Is Correct
TSH low ( < 0.3) (1 ~, FTI, 2 T FTI, 10 normal FTI)--
lO false-positive TSH values / 9.3%
11 false-negative FTI
TSH high ( > 5) (20 $ FTI, 2 1' FTI, 22 normal FTI)--
24 false-positive TSH values
J
If TSH Is Correct
FTI high 1' (2 I' TSH, 4 $ TSH, 7normalTSH)--
9false-positive FTIvalues
FTI low ( < 5) (2 $ TSH, 13 1' TSH, 26 "normal" TSH)--
28 false-positive FTI values
22 false-negative FTI If FTI Is Correct
l 10.2% J
9false-negative TSH 27 false-negative TSH
TSt-I = thyroid-stimulatinghormone;FTI = freethyroxineindex.
who had an elevated TSH but normal FTIs, and who were not in need of treatment. Assuming that one test is correct and the other possibly wrong, the incidence of false positives for the two tests can be calculated as shown in Table IX. The percent of false positives is effectively the same for each test. On the other hand, this analysis assumes that the low FTI in patients who have the ESS is a correct and meaningful observation, which is a matter of considerable discussion and no unanimity of opinion. A virtue, or problem, with the FTI determination is that it identifies a group of individuals who have a low FTI and a normal or suppressed TSH. These patients are presumed to have ESS. Current ideas on their proper management vary. Numerous studies on such patients suggest that, on analysis of serum free T4 by ultrafiltration or dialysis, 0% to 45% have low values, and the remainder normal or even elevated values of free hormone [9-11]. Individuals with ESS and an FTI or serum T4 value below 4 have a very poor prognosis. There is a direct correlation b e t w e e n the reduction in serum T4 or FTI and probability of fatal outcome of illness [12]. Two studies [13,14] have reported that administration of replacement hormone does not affect prognosis, although uncontrolled reports have suggested a benefit. No reports have suggested that replacement thyroid hormone treatment of patients with this condition adversely affects prognosis. While the present study does not bear on the appropriateness of therapy for this category of patients, it is clear that screening by FTI would raise this question in many patients. Our data indicate that screening by determination of serum T4 on all admissions to an acute care general teaching hospital would be inappropriate because of the significant number of patients incorrectly classified on the basis of this determination. Screening by TSH assay would identify patients
with hypothyroidism in need of thyroid hormone treatment, or those with overreplacement of T4. Screening by FTI would detect patients with hypothyroidism and those with excess levels of thyroid hormone, and in addition, would identify patients with ESS. All patients who had a significant abnormality by one test would presumably qualify for a follow-up determination of the alternate test, FTI or TSH. Selection of patients with age greater than 50 or by sex would miss many individuals with abnormal test results. Assuming that the screening TSH or FTI could be performed on a routine basis for approximately $10, and that a follow-up TSH or FTI determination is required on the approximately 10% of patients who are found to have an abnormality, at an average cost of $40, it can be calculated that the average increment in the hospital bill would be $14, if screening was carried out on all admissions. Attempts have been made to evaluate the costeffectiveness of screening determinations, but the assumptions, in terms of improved quality of life or decreased utilization of physician services, are problematic [6]. We do not think that it is currently possible to make such an analysis in a meaningful manner. Our data also indicate that patients found to require therapy would primarily be patients who had not been previously diagnosed. Whether they would have been correctly diagnosed independently by their physicians during hospitalization is uncertain, but appears unlikely. Only 3% of the patients recognized to have significantly abnormal FTI or TSH values had these tests ordered by their attending physicians prior to the availability of the results from the screening tests, suggesting that most of these diagnoses were not obvious and would not otherwise have come to attention. In community surveys, screening programs have detected a 0% to 1% incidence of undiagnosed hypo-
November 1992 The American Journal of Medicine Volume 93
563
SCREENING BY THYROID FUNCTION TESTS / DeGROOTAND MAYOR
thyroidism [5]. In studies carried out on hospitalbased populations, the incidence is considerably higher, ranging from 1% to 9% [1-5], depending upon the category of patient, and is typically highest among women over age 50. Helfand and Crapo [5] have reviewed the field carefully, and argue that case finding is a more appropriate and cost-effective approach than screening. Spencer et al [3] believe that screening by sensitive TSH assay is unsatisfactory, since many patients have low TSH values of uncertain significance, and indicate that screening with FTI is appropriate. Atkinson et al [1] concluded that screening by T4 of patients over age 60 was most appropriate, with follow-up by TSH and triiodothyronine, as needed. Livingston et al [2] recommend screening elderly inpatients with FTI and TSH. Ehrmann et al [15] reported that the sensitive TSH assay frequently provides misleading results in hospitalized patients. De los Santos et al [16] studied the utility of two radioimmunoassay methods (Abbott Laboratories, North Chicago, IL, and Serono Diagnostics, Inc., Norwell, MA) and the FTI, in detecting thyroid disease in 544 ambulatory subjects. While both TSH assays were more sensitive and specific, the authors concluded that the FTI was most cost-effective as the first-line test, but did not advocate routine screening of ambulatory patients. Small et al [17] conducted a study on hospital admissions very similar to ours and with very comparable results. Twenty percent of their patients had abnormal tests (24% in our data), 11.6% had ESS (we found 7.4%), and 6.2% had subclinical hypothyroidism (versus our 6.0%). They suggest screening of patients over age 60. Application of this age limit would have meant missing 58% of the individuals who had significantly abnormal test results in our study population, and hardly seems an appropriate selector. Our conclusion regarding the superiority of the FTI for screening, over the TSH or T4 assays, agrees with the majority of these reports. The incidence of apparent hypothyroidism detected in our stud y - a t least 5.8%--is higher than in many other reports. We believe our data are representative of the experience to be expected in this hospital, since our study used a large random sample. This incidence of hypothyroidism may also be typical of other large urban acute care general teaching hospitals. We believe that detection of a notable number of individuals with clinically significant hypothyroidism or elevated hormone levels, and possibly detection of ESS, has clinical value that offsets the modest average increment in the cost of hospitalization. Screening by TSH assay would identify up to 5.8%
564
November 1992
The American Journal of Medicine
of patients needing thyroid hormone replacement, but would also identify a group of patients who have subclinical hypothyroidism and others with suppressed TSH but normal FTI values that would require follow-up investigation, although the latter groups would not require therapy. Screening by FTI, which we favor, would identify the same group of patients with hypothyroidism (5.8%), and would immediately signal the severity of their thyroid hormone deficit. In addition, a larger group of patients (7.4%) who appear to have ESS would be identified. Whether these patients should or should not be treated is uncertain. Either test would require retesting a b o u t 10% of p a t i e n t s by follow-up examination.
REFERENCES 1. Atkinson RL, Dahms WT, Fisher DA, Nichols AL. Occult thyroid disease in an elderly hospitalized population. J Gerontol 1978; 33: 372-6. 2. Livingston EH, Hershman JM, Sawin CT, Yoshikawa 1-l. Prevalence of thyroid disease and abnormal thyroid tests in older hospitalized and ambulatory persons. J Am Geriatr Soc 1987; 35: 109-14. 3. Spencer C, ElgenA, Shen D, etal. Specificity of sensitive assays of thyrotropin (TSH) used to screen for thyroid disease in hospitalized patients. Clin Chem 1987; 33: 1391-6. 4. Eggertsen R, Petersen K, Lundberg P-A, NystrOm E, Lindstedt G. Screening for thyroid disease in a primary care unit with a thyroid stimulating hormone assay with a low detection limit. BMJ 1988; 297: 1586-92. 5. Helfand M, Crapo LM. Screening for thyroid disease. Ann Intern Med 1990; 112: 840-9. 6. Nolan JP, Tarsa NJ, DiBenedetto G. Case-finding for unsuspected thyroid disease: costs and health benefits. Am J Clin Pathol 1985; 83: 346-55. 7. Robin NI, Hagen SR, Collaco F, et al. Serum tests for measurement of thyroid function. Horm Res 1971; 2: 266-79. 8. Amino N, Hagen SR, Yamada N, Refetoff S. Measurement of circulating thyroid microsomal antibodies by the tanned red cell hemagglutination technique: its usefulness in the diagnosis of autoimmune thyroid diseases. Clin Endocrinol (Oxf) 1976; 5: 115-25. 9. DeGroot LJ. Abnormal thyroid function tests in euthyroid patients. In: Hamburger J, editor. Diagnostic methods in clinical thyroidology. New York: Springer-Verlag, 1989: 3-15. 10. Chopra I, Van Herle AJ, Chua Teco GN, Nguyen AH. Serum free thyroxine in thyroidal and nonthyroidal illnesses:a comparison of measurements by radioimmunoassay, equilibrium dialysis, and free thyroxine index. J Clin Endocrinol Metab 1980; 51: 135-43. 11. Wang Y-S, Pekary AE, England ML, Hershman JM. Comparison of a new ultrafiltration method for serum free T4 and free T3 with two RIA kits in eight groups of patients. J Endocrinol Invest 1985; 8: 495-9. 12. Wartofsky L, Surman KD. Alterations in thyroid function in sick patients with systemic illness: the "euthyroid sick syndrome." Endocr Rev 1982; 3: 164-217. 13. Brent GA, Hershman JM. Thyroxine therapy in patients with severe nonthyroidal illnesses and low serum thyroxine concentration. J Clin Endocrinol Metab 1986; 63: 1-8. 14. Becker RA, Vaughan GM, Ziegler MG, eta/. Hypermetabolic low triiodothyronine syndrome of burn injury. Crit Care Med 1982; 10: 870-5. 15. Ehrmann DA, Weinberg M, Same DH. Limitations to the use of a sensitive assay for serum thyrotropin in the assessment of thyroid status. Arch Intern Med 1989; 149: 369-72. 16. de los Santos ET, Starich GH, Mazzaferri EL. Sensitivity, specificity, and costeffectiveness of the sensitive thyrotropin assay in the diagnosis of thyroid disease in ambulatory patients. Arch Intern Med 1989; 149: 526-32. 17. Small M, Buchanan L, Evans R. Value of screening thyroid function in acute medical admissions to hospital. Clin Endocrinol (Oxf) 1990; 32: 185-91.
Volume 93
Chicago,///inois
To determine the incidence of unrecognized thyroid disease among admimaions to a large acute care university teachinr hospital, 364 sampies taken on consecutive admissions were assayed for thyroid-stlmulating hormone (TSH) and free thy ~ - e index (l~I). Patients with abnormnl test results were further evaluated by determin~tion of antlmlcresomal and antithyroglob-lin antibodies, and c h R m were reviewed for evidence of prior diagnosis of thyroid disease, especially severe illness, drug treatment that might affect thyroid function tests, and prior diagnosis of thyroid diseas~ Results of subsequent thyroid function tests performed during the patient's hospitali~tion were correlated with the admlmlon serum assays, and data on subsequent testing during the following 6 months were also obtained. A total of 3.9% of patients had significantly depressed TSH, and 11.1% of values were significantly elevated. A total of 11~1% of patients had significantly low FTI values, and 1% had significantly elevated values, A total of 7.4% appeared to have the euthyroid sick syndrome, 5~% appeared to have unrecognized or undertreated primRmy thyroid failure, 6% had apparent subclinical hypothyroidism, 2% were thyrotoxic, and 2~% (all women) had suppressed TSH levels for inapparent reasons. Limltin~ testing to patients over 49 years of age, or to women, would have mimed many individnAIA with abnormal test results. Considering widespread availability of tests, relative costs, and value of the information obtained, it is suggested thet the FTI determ;n=tion would provide an appropriate screening test for patients in a population such as this entering a large, acute care general hospital.
From the Thyroid Study Unit, Department of Medicine, The University of Chicago, Chicago, Illinois. This work was supported by United States Public Health Service Grants DK13377 and DK27384, the March of Dimes Birth Defects Foundation, Boots Pharmaceuticals Inc., and the David Wiener Research Fund. Requests for reprints should be addressedto Leslie J. DeGroot, M.D., Thyroid Study Unit/Mail Code 3090, The University of Chicago, 5841 South Maryland Avenue, Chicago, Illinois 60637. Manuscript submitted September 26, 1991, and accepted in revised form August 19, 1992.
558
he value of screening populations for undiagnosed thyroid disease has been investigated in T a variety of settings including elder-care institutions, Veterans Administration hospitals, acute general hospitals, and outpatient clinics [1-5]. Typically, 2% to 10% of subjects are found to have undiagnosed thyroid disease. While the medical benefits of detecting even mild thyroid disease are presumed to be self-evident, the analysis of economic benefit has been difficult but appears to favor screening [6]. We note that thyroid function tests are believed to already account for nearly a billion of the several hundred billion dollars spent on health care each year in the United States [6]. Nevertheless, some level of laboratory screening of new admissions to acute care hospitals is commonly performed. We reasoned that substitution of a relatively inexpensive thyroid function test in an admission "chemistry profile" might be cost-effective, and perhaps could replace some other component of the profde. For example, the serum phosphate, a typical component of the screening chemistry profile, rarely provides significant information, although it obviously may be of extreme importance in renal disease, ketoacidosis, and other disorders in which attention is specifically directed to serum phosphate determination. To explore this question further, we performed thyroid function tests on nearly 400 consecutive, unselected admissions to an acute care general teaching hospital, analyzed the relationship of the abnormalities discovered to characteristics of the population, and compared the utility of different assays for screening.
METHODS With the approval of the Institutional Review Board, discarded serum samples were obtained from the hospital chemistry laboratory. The first serum samples sent to the chemistry laboratory, within 24 hours of admission of a new patient, were collected until 400 consecutive samples had been recovered. For technical reasons (duplication, insufficient serum, contamination, inadequate identification, and so forth), some were excluded, leaving 364 samples available for analysis. On this group of sera, thyroid-stimulating hormone (TSH) was assayed using an immunoradioactive assay (Serono,
November 1992 The American Journal of Medicine Volume 93
SCREENING BY THYROID FUNCTION TESTS / DeGROOT AND MAYOR
TABLE I
TABLE II
Distribution of Total Population by TSH Values
Distribution of Total Population by FTI Values
TSH (I~U/mL)
Frequency
%
Age (y)* < 50 > 49
0-0.2 0.3 0.4-4 4.1-5 5.1-8 >8
14 9 280 17 24 20
3.9 2.5 76.8 4.7 6.6 5.5
5 7 130 7 7 3
Total
364
100.0
159
9 2 150 10 17 17 205
FTI (units)
Sex* M
F
0 3 136 4 7 5
14 6 144 13 17 15
155
209
TSH= thyroid-stimulatinghormone, *Distributionof resultsnonrandomin relationtoageandsex(p < 0,05).
49 16
25
Sex(NS) M F 20
21
5-5.9
52
14.3
27
25
23
29
6-10.5
258
70.8
110
148
108
150
10.6-11.5
9
2.5
4
5
3
6
>11.5
4
1.1
2
2
1
3
364
100.0
159
205
155
209
Total
FTI= freethyroxineindex;NS= distributionofpatientsbyageandsexis randomin relationto FTI,
Allentown, PA), thyroxine (T4) was assayed using an immunoassay (Diagnostic Products, Los Angeles, CA), and free thyroxine index (FTI) was derived as reported using a resin T4 uptake method [7]. In normal subjects studied in our laboratory, serum TSH levels display a bell-shaped distribution, with 98% of values ranging from 0.4 to 4 pU/mL, which was taken to be the normal range. Serum TSH levels were separated into six categories: 0 to 0.2 #U/mI~assumed to be significantly low; 0.3 pU/mL--considered borderline; 0.4 to 4 pU/mL--considered normal; 4.1 to 5 pU/mL--considered minimally elevated; 5.1 to 8 ~U/mL--considered elevated; and greater than 8 pU/mL--considered seriously elevated. T4 values were similarly categorized. Values less than 4 pg/dL were considered significantly low, 4.4 to 4.9 ~g/dL borderline, 5 to 12 pg/dL normal, 12.1 to 13 pg/dL minimally elevated, and greater than 13 ~g/dL significantly elevated. Serum FTI values were broken into corresponding categories: less than 5, indicating significant depression; 5 to 5.9, borderline; 6 to 10.5, normal; 10.6 to 11.5, borderline elevated; and greater than 11.5, significantly elevated. Age and sex of patients were recorded. Data from patients with significantly low or elevated FTI or TSH were further reviewed by analysis of laboratory records and hospital charts. The hospital course was analyzed to determine the presence of extremely severe illness on admission, use of medications that might depress or elevate the TSH or FTI (dopamine, prednisone, diphenylhydantoin, propylthiouracil, thyroxine), and previously or concurrently diagnosed thyroid disease. In this subgroup of patients, serum was also analyzed for the presence of antimicrosomal and antithyroglobulin antibodies by hemagglutination assay [8]. Results were tabulated in a "Dbase" (Ashton-Tate, Culver City, CA) program and further analyzed by program "Stata" (Computing Resource Center, Los Angeles,
CA). Group differences observed in linear regression or ×2 analysis were considered significant if the p value was 1/20. *Distributionof resultsrandomin relationto totalpatientgroup. tOistributionof resultsnonrandomin relationto totalpatientgroup.
TABLE IV Characteristics of Patients With Abnormal FTI or TSH, Grouped by FTI Values Age* FTI
Frequency
Sext
ESS*
%
> 49
< 50
M
F
+
-
0
TDX* 1
2
0
Drug* 1
2
Antibodies* +
11.5
41 7 32 3 4
47.1 8.1 36.7 3.5 4.6
25 6 24 1 2
16 1 8 2 2
20 5 3 1 1
21 2 29 2 3
23 4 13 0 1
18 3 19 3 3
37 6 25 2 1
4 1 6 1 2
0 0 1 0 1
33 6 21 2 2
5 1 6 0 1
3 0 5 1 1
30 3 18 3 1
9 3 10 0 1
Total
87
i00.0
58
29
30
57
41
46
71
14
2
64
13
10
55
23
FTI = freethyroxineindex;TSH = thyroid-stimulatinghormone;ESS= euthyroidsicksyndrome(verysevereillnessasestimatedby reviewof clinicalrecord);TDX= 1--prior diagnosisof Hashimoto's thyroiditisor hypothyroidism,or 2--prior diagnosisof hyperthyroidism;Drug= 1--prednisone,diphenylhydantoin,aspirin,dopamine,propylthiouraciladministration,or 2--thyroxineadministration; Antibodies= microsomalorthyroglobuiinhemagglutinationtiter > 1/20. *Distributionof resultsrandomin relationto totalpatientgroup. t Distributionof resultsnonrandomin relationtototalpatientgroup.
TABLE V Distribution of Total Population by TSH and FTI Values TSH (I~U/mL) 0-4.9
5-5.9
FTI (units) 6-10.5 ]0.6-11.5
>11.5
Total
0-0.2 0.3 0.4-4 4.1-5 5.1-8 >8
1' 1" 25* 111 211 1 llr
0 0§ 47§ 2§ 511 211
101 7§ 201§ 14§ 164 64
1 0§ 6§ 0§ 1 1
2' 1 1 0 0 0
14 9 280 17 24 20
Total
41
56
254
9
4
364
TSH= thyroid-stimulatinghormone;FTI = freethyroxineindex. *Euthyroidsicksyndrome(7.4%of patients). tSuppreesedTSH(2,8%of patients). tToxic(1,1%of patients). §Paranormal. IIHypothyroid(5.8%of patients). ISubclinicalhypothyroidism(6.0%of patients).
Characteristics of Patients With Either Abnormal TSH or FTI To further characterize the group of patients with abnormal TSH or FTI values, hospital records were evaluated for evidence of extremely severe ill560
ness, prior thyroid diagnosis, use of certain medications, or presence of autoimmune thyroid disease. As shown in Table III, patients with low TSH uniquely were female. Nearly half (6 of 14) had a previous diagnosis compatible with autoimrnune thyroid disease, and 4 of 15 were taking medications that might suppress TSH. Not surprisingly, antithyroid antibodies were detected in 10 of 19 patients with TSH elevated above 8 #U/mL. Analysis of age, sex, severity of illness, prior history of thyroid disease, medications, and antithyroid antibodies, in relation to serum FTI, indicated a striking excess of males among patients with a reduced FTI (Table IV). Categorization of the Population by TSH and FTI Values In theory, determination of TSH and FTI should correctly place most patients into diagnostic categories. This distribution is shown in Table V. Two hundred seventy-seven of the patients fell into a normal or "paranormal" category. Twenty-seven patients had a low FTI with TSH normal or sup-
November 1992 The American Journal of Medicine Volume 93
SCREENING BY THYROID FUNCTION TESTS / DeGROOT AND MAYOR TABLE VI Relation of Clinical Factors to Categorizationof Patients by FTI and TSH Values ApparentEuthyroid
ApparentPrimary
SickSyndrome
Hypothyroidism
Suppressed
TSH
Subclinical
Apparent
Total Group
Hypothyroidism Hyperthyroidism
Priorthyroiddiagnosis None Hashimoto'sthyroiditis Graves'
26 1 0
17 4 0
6 4 0
16 6 0
0 2 0
65 17 0
Severeillness + -
14 13
13 8
4 6
9 13
0 2
40 42
Medication None Category1 (TableIII) Category2 (TableIII)
22 4 1
17 2 2
4 3 3
17 3 2
1 1 0
61 13 8
Antibodies + -
3 23
9 10
4 5
6 13
1 0
23 51
Age 49
12 15
5 16
5 5
3 19
0 2
25 57
Sex M F
16 11
9 12
0 10
3 19
0 2
28 54
FTI = free thyroxine index; TSH = thyroid-stimulating hormone.
pressed, suggesting the "euthyroid sick syndrome" (ESS) or, much less likely, hypopituitarism. This group comprised 7.4% of the entire population. As shown in Table VI, there was an excess of males in the ESS group, and 14 of 27 had unusually severe illness, but the patients were not significantly older than the entire group, and only 4 were taking drugs that might significantly alter FTI. Twenty-one (5.8%) had low FTI and elevated TSH, indicating primary hypothyroidism (Table V). Four of these patients had previously been recognized to have thyroid disease, but had not been given replacement therapy, or had been given it in inadequate amounts. These patients were older than the average, more (62%) had severe illness, and a higher proportion (47%) had positive antibodies, fitting with a probable diagnosis of Hashimoto's thyroiditis (Table VI). A total of 2.8% of patients, all women, had a norreal FTI with suppressed TSH. In three, medication might have been responsible, but in the remainder the cause is obscure. Perhaps it is a form of ESS. Six percent of patients had an elevated TSH with normal FTI, suggesting subclinical hypothyroidism; these patients were primarily older women, and 6 of 16 had some known prior clinical or laboratory finding suggesting they had Hashlmoto's thyroiditis. There was no excess of positive antibody assays in this group. Two patients had an elevated FTI and suppressed TSH, suggesting hyperthyroidism. One patient had been given excess
TABLE VII
Distributionof Total Patients by SerumT4 Values T4
(iLg/dL)
Frequency
%
>4 4-4.9 5-12 12.1-13 > 13
31 28 286 9 10
8.5 7.7 78.5 2.5 2.8
Total
364
100.0
T4 = thyroxine.
thyroid hormone for replacement of primary thyroid failure (Tables V and VI). Five patients with abnormal testsdid not fitinto any obvious diagnostic pattern.
Analysis of Serum T4 Values Analysis of serum T4 values showed that 8.5% of patients had a value less than 4 pg/dL, and 2.8% had a value greater than 13 #g/dL. This distribution is very similar to that of the FTI values (Table VII). Patient categorization by TSH and serum T4 values was evaluated (data not shown). While the distribution of patients in terms of numbers in each category was quite similar to that observed for FTI, on analysis it was found that the patients in the categories differed. For example, 11 of the patients classified as having ESS, on the basis of FTI and TSH,
November 1992 The American Journal of Medicine Volume 93
561
SCREENING BY THYROID FUNCTION TESTS / DeGROOT AND MAYOR
total population (including 45% of those with abnormal test results), another thyroid function test was ordered within 6 months. Obviously, other tests were probably performed in other laboratories as well.
TABLE VIII Clinical Review of Patients With Abnormal TSH TSH 8 TSH = thryoid-stimulatinghormone;T,L= thyroxine;PTU = propylthiouracil;FTI = fleethyroxine index;Rx= therapy,
were not present in the similar group classified on the basis of s e r u m T4. Five of the patients classified as having primary thyroid failure by FTI in Table V did not appear in this category when analyzed on the basis of serum TSH and serum T4. FTI is well known to be a better index of available thyroid hormone than serum T4. Because of this background and the apparent errors in categorization on the basis of s e r u m T4, further analysis of patients on the basis of s e r u m T4 values was not performed. Diagnosis and Follow-Up All seriously abnormal results were reported directly to the patient's physician, and reports were posted in the patients' charts. By analysis of laboratory records, it was found that in only 1% of the entire population (3% of the patients with abnormal tests) had the attending physicians ordered thyroid function tests prior to availability of the results from this survey. Thirty-seven patients had a second FTI determination following report Of the first sample. The correlation coefficient of the two assays was 0.74. In 11% the new FTI value removed the patient from the original diagnostic category. Thirty-nine patients (including 44% of those with an abnormal test result) had a second in-hospital TSH determination. The correlation coefficient was lower, 0.54, often because the second TSH leveis were much more elevated. In 13%, the .result shifted the patient to a different diagnostic category. Considering that some patients received specific treatment or had altered clinical conditions or treatments between tests, it is apparent that repetition verified the original observation in nearly 90% of the test cases. We also found that in 11.3% of the 562
November1992 The American J o u m ~ of Medicine
On the basis of assay of serum TSI-I and FTI, 6% of patients were found to have subclinical hypothyroidism, 5.8% to have probable primary hypothyroidism, and 3% to have an FTI less than 5 and TSH greater than 8, indicating an urgent need for thyroid hormone replacement therapy (Table I, Table V, and Table VIII). Overall, 20 of the 364 patients, or 5.5%, needed thyroid hormone replacement therapy, as judged by usual standards of medical management. Of the 13 patients with a significantly suppressed TSH, the cause was due to overreplacement with thyroid hormone in 3 and probably the ESS in 1, but the cause was inapparent in the majority of cases who were older women. Serum TSH assay identified as abnormal individuals with apparent primary hypothyroidism, mainly individuals without a known prior diagnosis, and a significant number of patients (20 patients, 5.5%) who might benefit from treatment. While this population had an increased proportion of individuals who were over age 50, and female, limiting testing to women above age 50 would have missed others: males (5 of 20) and younger patients (3 of 20) with significant hypothyroidism. Three of 15 patients with suppressed TSH appeared to have this finding on the basis of administered thyroxine. Screening by TSH assay also led to identification as abnormal a significant number (6%) of patients with subclinical hypothyroidism, and 2.8% with only suppressed TSH. These patients presumably do not need therapy at this time for a thyroid problem. Screening on the basis of FTI indicated that 11.3% had significantly low values and 1.1% had elevated values. The low FTI values were Rimost equally divided between patients who appeared to have ESS and patients with apparent primary hypothyroidism. FTI did not, by definition, identify patients with "subclinical hypothyroidism." Patients with abnormal FTIs were not clustered on the basis of age or sex. Drug effects did not explain a significant proportion of the abnormal results, and, somewhat surprisingly, there was no excess of extremely ill patients among the individuals with marked depression of FTI. Clearly, neither TSH nor FTI screening alone properly identified all patients with a need for therapy. For example, while TSH identified the individuals with apparent primary hypothyroidism, it also identified a group of individuals, equal in number,
Volume 93
SCREENING BY THYROID FUNCTION TESTS / DeGROOT AND MAYOR
TABLE IX Analysisof TSH/FTI Data If FTI Is Correct
If TSH Is Correct
TSH low ( < 0.3) (1 ~, FTI, 2 T FTI, 10 normal FTI)--
lO false-positive TSH values / 9.3%
11 false-negative FTI
TSH high ( > 5) (20 $ FTI, 2 1' FTI, 22 normal FTI)--
24 false-positive TSH values
J
If TSH Is Correct
FTI high 1' (2 I' TSH, 4 $ TSH, 7normalTSH)--
9false-positive FTIvalues
FTI low ( < 5) (2 $ TSH, 13 1' TSH, 26 "normal" TSH)--
28 false-positive FTI values
22 false-negative FTI If FTI Is Correct
l 10.2% J
9false-negative TSH 27 false-negative TSH
TSt-I = thyroid-stimulatinghormone;FTI = freethyroxineindex.
who had an elevated TSH but normal FTIs, and who were not in need of treatment. Assuming that one test is correct and the other possibly wrong, the incidence of false positives for the two tests can be calculated as shown in Table IX. The percent of false positives is effectively the same for each test. On the other hand, this analysis assumes that the low FTI in patients who have the ESS is a correct and meaningful observation, which is a matter of considerable discussion and no unanimity of opinion. A virtue, or problem, with the FTI determination is that it identifies a group of individuals who have a low FTI and a normal or suppressed TSH. These patients are presumed to have ESS. Current ideas on their proper management vary. Numerous studies on such patients suggest that, on analysis of serum free T4 by ultrafiltration or dialysis, 0% to 45% have low values, and the remainder normal or even elevated values of free hormone [9-11]. Individuals with ESS and an FTI or serum T4 value below 4 have a very poor prognosis. There is a direct correlation b e t w e e n the reduction in serum T4 or FTI and probability of fatal outcome of illness [12]. Two studies [13,14] have reported that administration of replacement hormone does not affect prognosis, although uncontrolled reports have suggested a benefit. No reports have suggested that replacement thyroid hormone treatment of patients with this condition adversely affects prognosis. While the present study does not bear on the appropriateness of therapy for this category of patients, it is clear that screening by FTI would raise this question in many patients. Our data indicate that screening by determination of serum T4 on all admissions to an acute care general teaching hospital would be inappropriate because of the significant number of patients incorrectly classified on the basis of this determination. Screening by TSH assay would identify patients
with hypothyroidism in need of thyroid hormone treatment, or those with overreplacement of T4. Screening by FTI would detect patients with hypothyroidism and those with excess levels of thyroid hormone, and in addition, would identify patients with ESS. All patients who had a significant abnormality by one test would presumably qualify for a follow-up determination of the alternate test, FTI or TSH. Selection of patients with age greater than 50 or by sex would miss many individuals with abnormal test results. Assuming that the screening TSH or FTI could be performed on a routine basis for approximately $10, and that a follow-up TSH or FTI determination is required on the approximately 10% of patients who are found to have an abnormality, at an average cost of $40, it can be calculated that the average increment in the hospital bill would be $14, if screening was carried out on all admissions. Attempts have been made to evaluate the costeffectiveness of screening determinations, but the assumptions, in terms of improved quality of life or decreased utilization of physician services, are problematic [6]. We do not think that it is currently possible to make such an analysis in a meaningful manner. Our data also indicate that patients found to require therapy would primarily be patients who had not been previously diagnosed. Whether they would have been correctly diagnosed independently by their physicians during hospitalization is uncertain, but appears unlikely. Only 3% of the patients recognized to have significantly abnormal FTI or TSH values had these tests ordered by their attending physicians prior to the availability of the results from the screening tests, suggesting that most of these diagnoses were not obvious and would not otherwise have come to attention. In community surveys, screening programs have detected a 0% to 1% incidence of undiagnosed hypo-
November 1992 The American Journal of Medicine Volume 93
563
SCREENING BY THYROID FUNCTION TESTS / DeGROOTAND MAYOR
thyroidism [5]. In studies carried out on hospitalbased populations, the incidence is considerably higher, ranging from 1% to 9% [1-5], depending upon the category of patient, and is typically highest among women over age 50. Helfand and Crapo [5] have reviewed the field carefully, and argue that case finding is a more appropriate and cost-effective approach than screening. Spencer et al [3] believe that screening by sensitive TSH assay is unsatisfactory, since many patients have low TSH values of uncertain significance, and indicate that screening with FTI is appropriate. Atkinson et al [1] concluded that screening by T4 of patients over age 60 was most appropriate, with follow-up by TSH and triiodothyronine, as needed. Livingston et al [2] recommend screening elderly inpatients with FTI and TSH. Ehrmann et al [15] reported that the sensitive TSH assay frequently provides misleading results in hospitalized patients. De los Santos et al [16] studied the utility of two radioimmunoassay methods (Abbott Laboratories, North Chicago, IL, and Serono Diagnostics, Inc., Norwell, MA) and the FTI, in detecting thyroid disease in 544 ambulatory subjects. While both TSH assays were more sensitive and specific, the authors concluded that the FTI was most cost-effective as the first-line test, but did not advocate routine screening of ambulatory patients. Small et al [17] conducted a study on hospital admissions very similar to ours and with very comparable results. Twenty percent of their patients had abnormal tests (24% in our data), 11.6% had ESS (we found 7.4%), and 6.2% had subclinical hypothyroidism (versus our 6.0%). They suggest screening of patients over age 60. Application of this age limit would have meant missing 58% of the individuals who had significantly abnormal test results in our study population, and hardly seems an appropriate selector. Our conclusion regarding the superiority of the FTI for screening, over the TSH or T4 assays, agrees with the majority of these reports. The incidence of apparent hypothyroidism detected in our stud y - a t least 5.8%--is higher than in many other reports. We believe our data are representative of the experience to be expected in this hospital, since our study used a large random sample. This incidence of hypothyroidism may also be typical of other large urban acute care general teaching hospitals. We believe that detection of a notable number of individuals with clinically significant hypothyroidism or elevated hormone levels, and possibly detection of ESS, has clinical value that offsets the modest average increment in the cost of hospitalization. Screening by TSH assay would identify up to 5.8%
564
November 1992
The American Journal of Medicine
of patients needing thyroid hormone replacement, but would also identify a group of patients who have subclinical hypothyroidism and others with suppressed TSH but normal FTI values that would require follow-up investigation, although the latter groups would not require therapy. Screening by FTI, which we favor, would identify the same group of patients with hypothyroidism (5.8%), and would immediately signal the severity of their thyroid hormone deficit. In addition, a larger group of patients (7.4%) who appear to have ESS would be identified. Whether these patients should or should not be treated is uncertain. Either test would require retesting a b o u t 10% of p a t i e n t s by follow-up examination.
REFERENCES 1. Atkinson RL, Dahms WT, Fisher DA, Nichols AL. Occult thyroid disease in an elderly hospitalized population. J Gerontol 1978; 33: 372-6. 2. Livingston EH, Hershman JM, Sawin CT, Yoshikawa 1-l. Prevalence of thyroid disease and abnormal thyroid tests in older hospitalized and ambulatory persons. J Am Geriatr Soc 1987; 35: 109-14. 3. Spencer C, ElgenA, Shen D, etal. Specificity of sensitive assays of thyrotropin (TSH) used to screen for thyroid disease in hospitalized patients. Clin Chem 1987; 33: 1391-6. 4. Eggertsen R, Petersen K, Lundberg P-A, NystrOm E, Lindstedt G. Screening for thyroid disease in a primary care unit with a thyroid stimulating hormone assay with a low detection limit. BMJ 1988; 297: 1586-92. 5. Helfand M, Crapo LM. Screening for thyroid disease. Ann Intern Med 1990; 112: 840-9. 6. Nolan JP, Tarsa NJ, DiBenedetto G. Case-finding for unsuspected thyroid disease: costs and health benefits. Am J Clin Pathol 1985; 83: 346-55. 7. Robin NI, Hagen SR, Collaco F, et al. Serum tests for measurement of thyroid function. Horm Res 1971; 2: 266-79. 8. Amino N, Hagen SR, Yamada N, Refetoff S. Measurement of circulating thyroid microsomal antibodies by the tanned red cell hemagglutination technique: its usefulness in the diagnosis of autoimmune thyroid diseases. Clin Endocrinol (Oxf) 1976; 5: 115-25. 9. DeGroot LJ. Abnormal thyroid function tests in euthyroid patients. In: Hamburger J, editor. Diagnostic methods in clinical thyroidology. New York: Springer-Verlag, 1989: 3-15. 10. Chopra I, Van Herle AJ, Chua Teco GN, Nguyen AH. Serum free thyroxine in thyroidal and nonthyroidal illnesses:a comparison of measurements by radioimmunoassay, equilibrium dialysis, and free thyroxine index. J Clin Endocrinol Metab 1980; 51: 135-43. 11. Wang Y-S, Pekary AE, England ML, Hershman JM. Comparison of a new ultrafiltration method for serum free T4 and free T3 with two RIA kits in eight groups of patients. J Endocrinol Invest 1985; 8: 495-9. 12. Wartofsky L, Surman KD. Alterations in thyroid function in sick patients with systemic illness: the "euthyroid sick syndrome." Endocr Rev 1982; 3: 164-217. 13. Brent GA, Hershman JM. Thyroxine therapy in patients with severe nonthyroidal illnesses and low serum thyroxine concentration. J Clin Endocrinol Metab 1986; 63: 1-8. 14. Becker RA, Vaughan GM, Ziegler MG, eta/. Hypermetabolic low triiodothyronine syndrome of burn injury. Crit Care Med 1982; 10: 870-5. 15. Ehrmann DA, Weinberg M, Same DH. Limitations to the use of a sensitive assay for serum thyrotropin in the assessment of thyroid status. Arch Intern Med 1989; 149: 369-72. 16. de los Santos ET, Starich GH, Mazzaferri EL. Sensitivity, specificity, and costeffectiveness of the sensitive thyrotropin assay in the diagnosis of thyroid disease in ambulatory patients. Arch Intern Med 1989; 149: 526-32. 17. Small M, Buchanan L, Evans R. Value of screening thyroid function in acute medical admissions to hospital. Clin Endocrinol (Oxf) 1990; 32: 185-91.
Volume 93
![[Hospital admission: An opportunity for diabetes screening].](/assets/img/hold.png)
