Background
Subclinical hypothyroidism (SCHo) is a biochemical diagnosis characterised by a raised serum thyrotrophin concentration (thyroid stimulating hormone (TSH)) accompanied by a normal concentration of free thyroxine (FT4) and is found frequently in older age (Wiersinga, Reference Wiersinga1995; Ross, Reference Ross2001; Wilson et al., Reference Wilson, Parle, Roberts, Roalfe, Hobbs, Clark, Sheppard, Pattison and Franklyn2006). Recent estimates suggest a UK prevalence of 2.6%–3.1% in individuals aged 65 years and above (Wilson et al., Reference Wilson, Parle, Roberts, Roalfe, Hobbs, Clark, Sheppard, Pattison and Franklyn2006). Diagnostic rates are unknown and are likely to rise due to both an aging population and an increasing use of thyroid function tests (TFTs) to rule out thyroid dysfunction, with recent data showing that 10 million TFTs are performed each year in the United Kingdom with the majority of requests being generated in primary care (Association for Clinical Biochemistry, British Thyroid Association and British Thyroid Foundation, 2006).
The clinical manifestation and consequences of overt hypothyroidism are well established and can be successfully treated through administration of thyroxine replacement therapy. In contrast, there is a lack of consensus with respect to the pathophysiological consequences of SCHo particularly in older individuals. Although SCHo is essentially a biochemical diagnosis, some studies have reported a greater prevalence of symptoms in patients with SCHo compared with age matched euthyroid controls (Zulewski et al., Reference Zulewski, Muller, Exer, Miserez and Staub1997; Canaris et al., Reference Canaris, Manowitz, Mayor and Ridgway2000; Meier et al., Reference Meier, Staub, Roth, Guglielmetti, Kunz, Miserez, Drewe, Huber, Herzog and Muller2001).
Others have suggested an association between SCHo and raised serum lipids (Danese et al., Reference Danese, Ladenson, Meinert and Powe2000), cardiac dysfunction (Biondi et al., Reference Biondi, Fazio, Palmieri, Carella, Panza, Cittadini, Bone, Lombardi and Sacca1999; Monzani et al., Reference Monzani, Di Bello, Caraccio, Bertini, Giorgi, Giusti and Ferrannini2001) and increased cardiac and all-cause mortality (Rodondi et al., Reference Rodondi, den Elzen, Bauer, Cappola, Razvi, Walsh, Asvold, Iervasi, Imaizumi, Collet, Bremner, Maisonneuve, Sgarbi, Khaw, Vanderpump, Newman, Cornuz, Franklyn, Westendorp, Vittinghoff and Gussekloo2010). However, other data show that raised TSH concentration confers a survival benefit in octogenarians (Gussekloo et al., Reference Gussekloo, van Exel, de Craen, Meinders, Frolich and Westendorp2004). The main argument advanced for treatment of SCHo is the risk of progression to overt hypothyroidism with progression rates varying from 2% to 17% (Gordin and Lamberg, Reference Gordin and Lamberg1981; Nystrom et al., Reference Nystrom, Caidahl, Fager, Wikkelso, Lundberg and Lindstedt1988; Parle et al., Reference Parle, Franklyn, Cross, Jones and Sheppard1993; Vanderpump et al., Reference Vanderpump, Tunbridge, French, Appleton, Bates, Clark, Grimely Evans, Hasan, Rodgers and Tunbridge1995). Risk of progression is further enhanced in the presence of antithyroid antibodies and with increasing serum thyrotrophin concentration (Surks et al., Reference Surks, Ortiz, Daniels, Sawin, Col, Cobin, Franklyn, Hershman, Burman, Denke, Gorman, Cooper and Weissman2004). With respect to treatment of SCHo, trials have generally been small, with some reporting an improvement in symptoms and others failing to demonstrate any clinical benefits (Cooper et al., Reference Cooper, Halpern, Wood, Levin and Ridgway1984; Nystrom et al., Reference Nystrom, Caidahl, Fager, Wikkelso, Lundberg and Lindstedt1988; Jaeschke et al., Reference Jaeschke, Guyatt, Gerstein, Patterson, Molloy, Cook, Harper, Griffith and Carbotte1996; Meier et al., Reference Meier, Staub, Roth, Guglielmetti, Kunz, Miserez, Drewe, Huber, Herzog and Muller2001; Razvi et al., Reference Razvi, Ingoe, Keeka, Oates, McMillan and Weaver2007; Parle et al., Reference Parle, Roberts, Wilson, Pattison, Roalfe, Haque, Heath, Sheppard, Franklyn and Hobbs2010). In older individuals the benefits of treatment must be balanced against the additional risks associated with polypharmacy.
Overall, evidence relating to SCHo remains inconclusive and deemed insufficient to justify population screening or routine treatment. Recommendations and guideline documents pertaining to management of SCHo are inconsistent, being based upon consensus opinion (see Table 1; Vanderpump et al., Reference Vanderpump, Ahlquist, Franklyn and Clayton1996; Adlin, Reference Adlin1998; Surks et al., Reference Surks, Ortiz, Daniels, Sawin, Col, Cobin, Franklyn, Hershman, Burman, Denke, Gorman, Cooper and Weissman2004; Association for Clinical Biochemistry, British Thyroid Association and British Thyroid Foundation, 2006; Mayo Clinic, 2007; Villar et al., Reference Villar, Saconato, Valente and Atallah2007; NHS Clinical Knowledge Summaries, 2010). Few data are available on how these recommendations are taken up in routine care or how SCHo is managed, and the aim of this study is therefore to describe these.
Table 1 Summary of current guidelines
ATA = American Thyroid Association; AACE = American Association of Clinical Endocrinologists; RCP = Royal College Physicians; DM = Diabetes Mellitus; ACP = American College of Physicians; NHS = National Health Service; TSH = thyroid stimulating hormone; FT4 = free thyroxine.
Methods
In 2002–2003, the Birmingham Elderly Thyroid Study (BETS) recruited 5881 individuals aged 65 years or more from 20 practices in central England. A few patients were excluded on the grounds of lack of capacity or serious illness, but in all other respects the cohort represents a community dwelling elderly population with 68% of invited individuals participating. Full details of the BETS cohort have been reported elsewhere (Wilson et al., Reference Wilson, Parle, Roberts, Roalfe, Hobbs, Clark, Sheppard, Pattison and Franklyn2006).
Follow-up and re-screening of thyroid function in this cohort (BETS2), excluding patients with overt thyroid dysfunction, was undertaken in 19 of the original 20 practices after an interval period of approximately five years. Overall, 3005 individuals from the baseline cohort participated in follow-up (BETS2).
Thyroid function was defined by the TFT results at re-screen or at initiation of thyroxine replacement therapy where this had occurred in the interval period between BETS1 and BETS2. TFTs were extracted from GP records and tests may have been performed in a variety of laboratories. Thyroid function was categorised as overt hyperthyroid, subclinical hyperthyroid, euthyroid, subclinical hypothyroid or overt hypothyroid in accordance with standard reference criteria of the main laboratory serving this GP area. Classification was based purely on TFT result and did not include consideration of symptoms. All BETS2 participants who had commenced thyroxine replacement therapy following a SCHo test result (defined by a TSH concentration above reference range and FT4 within reference range), in the period between screening for BETS1 and BETS2, were eligible for inclusion in the current study.
Review of guideline documents and statements of expert opinion informed development of the case report form and defined data to be extracted from the primary care records. The data collection period was defined as the interval between baseline thyroid function screening (BETS1) and commencement of treatment. All consultations during this interval period were reviewed and data extracted. Data extraction focused upon 12 symptoms suggestive of overt hypothyroidism (muscle weakness, muscle cramps, weight gain, cold intolerance, constipation, lethargy, cognitive impairment, dry skin, poor memory, hoarse voice, deep voice and facial oedema) and nine other thyroid relevant patient factors or clinical characteristics (goitre, TSH concentration >10 mI/UL, depression, abnormal lipid concentrations, advancing age, high cardiovascular risk score, family history of hypothyroidism, positive test for antithyroid antibodies and patient requested TFT). TFT dates, results and entire records of the consultation that triggered requests for TFT and each subsequent discussion related to initiation of replacement therapy were extracted.
Consultations were classified as ‘thyroid relevant’ or not according to the reason for the consultation. Thyroid relevant consultations were further categorised on the basis of two key decisions (i) to perform a TFT or (ii) to commence thyroxine replacement therapy. Patient demographics, including age, gender and index of multiple deprivation (IMD score 2004; a measure of deprivation; Noble et al., Reference Noble, Wright and Dibben2010) were obtained from the original study database.
In addition to case note review, one GP (the research lead) in each of the participating BETS practices was invited to complete a study questionnaire. The GP questionnaire was devised specifically to ascertain whether differences in practice policies or routine procedures existed and to identify which symptoms and/or other patient characteristics were perceived to influence clinical decision making with respect to management of SCHo in older individuals. The questionnaire consisted of 10 questions with categorical options and a free text section for comments.
Statistical analysis
Data were entered into a Microsoft Access database and analysed using the Statistical Packages for Social Sciences (SPSS-17) Windows package. All free text was reviewed and coded to enable numerical content analysis. Data were summarised using means with standard deviation (SD) and corresponding range for data that were normally distributed and median with inter-quartile range (IQR) for data that were not normally distributed to describe current practice.
Results
Of the 3005 individuals followed up, 42 had commenced thyroxine replacement therapy following an earlier test result indicating SCHo and were therefore eligible for inclusion in the current study (see Figure 1). These patients were registered at 14 of the 19 participating general practices with three practices accounting for 55% of such cases. Most (71.4%) were female and the mean age was 76.5 years (SD 5.49, range 66.8–88.3). Deprivation scores (IMD 2004 score) ranged from 3.8 (least deprived) to 60.8 (most deprived) with a median score of 14.6 (IQR 22.16).
Figure 1 Consort diagram defining study population. BETS = Birmingham Elderly Thyroid Study.
The mean interval period between baseline thyroid function screening (BETS1) and initiation of thyroxine replacement therapy was 2.52 years (range 0.17–4.45 years), giving a total follow-up of 105.8 patient years. During the interval period between baseline screening and commencement of therapy, a total of 854 consultations were recorded for these individuals; 17.4% (n = 149) of these consultations were deemed to be ‘thyroid relevant’, with a median number of 3.0 (IQR 3, range 1–14) thyroid relevant consultations being recorded per patient. Overall, 119 tests were undertaken between baseline screen and initiation of thyroxine therapy with 59 of these being repeat tests after an initial subclinical result had been received by the practice, equating to 1.4 confirmatory tests per patient before initiation of treatment.
Requests for thyroid function testing
Indications for the index TFT (the test that first indicated SCHo) were identified and recorded (see Table 2). The BETS baseline screening TFT was the index TFT for eight individuals and no justification for conducting the index TFT was recorded for 4/42 patients (Table 2).
Table 2 Documented reasons for performing TFTs and initiation of thyroxine replacement therapy
TFT = thyroid function test; BETS = Birmingham Elderly Thyroid Study; TSH = thyroid stimulating hormone; FT4 = free thyroxine.
Repeat TFTs were requested with a median interval of 102 days (IQR 157.5, range 4–803) following a result indicating euthyroid function and 96 days (IQR 136.5, range 4–412) after a result indicating SCHo. Documented reasons for repeating TFTs are presented in Table 2.
Initiation of thyroxine replacement therapy
Thyroxine replacement therapy was initiated in 32 (76%) patients following two TFTs confirming SCHo. In 10 patients, replacement therapy was initiated on the basis of a single TFT. TSH concentrations at the point of initiation of therapy are presented in Figure 2. Median TSH was 7.1 mIU/L (IQR 3.25, range 4.55–79.7) and median FT4 was 12.4 pmol/L (IQR 3.38, range 10.0–20.3). Reasons for initiation of thyroxine therapy were only documented for 26 of the 42 patients (61.9%) and are presented in Table 2. Prior to initiation of thyroxine replacement therapy measurement of antithyroid antibody was undertaken for six patients (five having a positive antibody titre).
Figure 2 Serum thyrotrophin concentrations immediately prior to initiation of thyroxine replacement therapy. TSH = thyroid stimulating hormone.
GP questionnaire
Overall 15/19 (79%) GPs representing 15 practices, completed and returned the study questionnaire. Two (13%) GPs reported having practice guidelines for management of SCHo. Most GPs (67%; 10/15), however, reported discussing management strategies with colleagues, with 6/15 (40%) reporting having general discussion to clarify usual practice and/or regarding specific patients.
Of the 12 symptoms suggestive of overt hypothyroidism that could trigger a request for TFT or influence the decision to initiate thyroxine, 91% of GPs stated that presence of cold intolerance and 82% that cognitive dysfunction or weight gain would trigger a TFT request (Table 3). Combinations of symptoms increased likelihood of GPs requesting a TFT and are presented in Table 3. One GP reported that TFT was routinely conducted in patients aged 65 years of age or more. When asked about their prescribing practice, 12 (80%) GPs reported ‘sometimes’ prescribing thyroxine replacement therapy for SCHo. Three responded that they had ‘never’ initiated treatment for SCHo, however, three GPs (20%) reported being aware of other GPs who routinely treated this borderline state (see Table 3).
Table 3 GP responses to questions relating to symptoms influencing management decisions
TFT = thyroid function test.
Twelve (92%) GPs reported that a serum TSH concentration >10 mIU/L would influence their decision to commence treatment for SCHo. The presence of antithyroid antibodies (77%, n = 10), raised serum lipid concentrations (62%, n = 8), goitre (39%, n = 5), family history (39%, n = 5) and depression (31%, n = 4) were also factors reported to be likely to effect GP decision making. Seven GPs reported that patient requests for treatment were important considerations and two suggested that decision making is influenced by age >80 years and presence of cardiovascular disease.
Discussion
Summary of the main findings
This study reviewed the routine primary care records of 42 elderly patients prescribed thyroxine replacement therapy in response to a TFT result suggestive of SCHo to identify all thyroid relevant consultations and examine indications for TFT and initiation of treatment. Eight hundred and fifty-four individual consultations were reviewed to capture all relevant data. A survey was also conducted to further explore GP rationale for diagnosing and managing SCHo in patients aged 65 years or more.
The study showed considerable variation in routine practice and between practices in the way GPs manage elderly patients with SCHo, with five practices not having initiated treatment on the basis of a subclinical result for any patient within the cohort and three practices accounting for 55% (23/42) of treated cases. This study is unable to comment on whether symptom burdens do differ between practices and this may in part account for the variation in practice. Importantly, GP responses suggest that GPs are uncertain how to manage SCHo in older individuals and that management strategies are largely intuitive due to a lack of clear evidence-based guidance to support decision making, highlighting an important training need.
In the current study, opportunistic thyroid function testing, undertaken during routine chronic disease clinics or due to presence of no specific symptoms lead to diagnosis of SCHo in more than 50% of cases. In only 12% of cases, the index TFT diagnosing SCHo was triggered by the presence of symptoms suggestive of hypothyroidism. Following this biochemical diagnosis, GPs used a combination of monitoring and treatment to manage SCHo. TFTs were frequently and repeatedly requested during the interval period between initial diagnosis and initiation of treatment with an average of 1.4 additional TFTs being conducted per patient during a mean interval period of 2.5 years. Repeated tests were triggered by a variety of patient factors, with monitoring of SCHo being the reason for 56% of the TFTs performed and presentation of symptoms. In terms of frequency of TFTs, shorter interval periods between TFTs were observed following results indicating higher serum concentration of thyrotrophin compared with results demonstrating borderline concentrations.
These findings are consistent with data reported in other studies. Following examination of data from a large healthcare database, Meyerovitch et al. reported an average frequency of three GP requested TFTs per person during a period of five years with an increased rate of testing with advancing age (Meyerovitch et al., Reference Meyerovitch, Rotman-Pikielny, Sherf, Battat, Levy and Surks2007). In a similar study, a range of between 0 and 5 requests for TFTs were observed over a period of 12 months in individuals with SCHo (Gibbons et al., Reference Gibbons, Conaglen and Lawrenson2010). This is in line with the median repeat interval of 96 days after an initial subclinical result observed in this study.
Few GPs in the current study reported using practice guidelines for routine management of SCHo. Most reported that their decision to initiate thyroxine replacement therapy was made on a case-by-case basis, driven predominantly by presence of symptoms suggestive of overt hypothyroidism. However, on the basis of the case review, only 12% of patients had thyroid symptoms recorded prior to the initial TFT. Unsurprisingly given the lack of consistency in guideline and recommendation papers, GP's decision making in this study appears to reflect a lower threshold for treating than intended by guidelines, for example, in 81% of cases GPs chose to initiate thyroxine therapy based upon a serum thyrotrophin concentration below 10 mIU/L and treatment was frequently initiated on the basis of a single subclinical result (24%). Findings from focus group research in New Zealand suggests that in the absence of guidelines or evidence to direct testing and treatment, GPs management decisions were based on a host of factors, including patient wishes, maintenance of a good doctor–patient relationship and medicolegal issues, and were not always based on symptom presentations or other clinical factors (Gibbons et al., Reference Gibbons, Lillis, Conaglen and Lawrenson2009). Data from the current study demonstrate that UK GPs adopt a similar non-evidence-based approach to management of SCHo in elderly individuals.
Limitations
The primary limitation of this study is the reliance upon retrospective routine clinical data recorded in the patient's medical records (Mistry et al., Reference Mistry, Buxton, Longworth, Chatwin and Peveler2005). Full data capture for some patients was not possible because the reasons for the test request or initiation of treatment were not documented. Furthermore, data may be inaccurate and/or incomplete due to documentation of only those factors considered relevant to the investigation or treatment undertaken. Inclusion of the GP questionnaire however enhanced data collection and corroborated findings of the case note review.
This study did not aim to systematically identify why GPs treat some patients with SCHo and not others, but it is likely that recording in consultations where treatment is initiated differs from recording in consultations where prescribing does not occur. Bias introduced by such a retrospective comparison is likely to have restricted the value of a control group. Identification of all individuals who had TFTs in the interval period could have enabled further consideration of this issue. This does not detract from the main findings of significant practitioner variability, little adherence to published recommendations in determining treatment decisions and a clear call from GPs for unified and consistent guidelines.
Routine practice with respect to TFT and management of SCHo within these 19 practices may be atypical as all participating practices had previously taken part in a larger programme of research exploring thyroid dysfunction in the elderly and therefore had interest in the research topic. The significant variation observed between practices suggest this is unlikely to be the case, but arguably interested practices are likely to be demonstrating more informed practice and the observed lack of adherence to recommendations is likely to be even more evident elsewhere. Follow-up of this cohort was not planned when the initial 2002–2003 study was undertaken. We therefore feel that management of patients in the interval period has been conducted in a largely naturalistic setting and the variation between practices is a real observation.
Conclusion
GPs are uncertain how to manage SCHo in older individuals and strategies for management of SCHo in the community therefore vary. Repeat tests are conducted with varying frequency, which is not informed by evidence, and thyroxine is prescribed for varying reasons with limited scientific rationale. These findings further demonstrate a clear need for definitive evidence-based guidelines to support GP decision making and to standardise management of SCHo in the community.
Acknowledgements
The authors would like to thank the patients and GPs who participated and also the general practice staff who facilitated collection of data. The co-investigators for BETS1 and/or BETS2 are Prof. Jayne Franklyn, Dr Michael Gammage, Dr Sayeed Haque, Mr Roger Holder, Prof. James Parle, Dr Helen Pattison, Mrs Andrea Roalfe and Prof. Sue Wilson. The study was designed by L.M.R., J.A. and DM and funding was secured by L.M.R. and F.D.R.H. J.A. and D.M. undertook management of the study, including overseeing data collection and management and quality assurance. J.A. undertook all analyses. All authors contributed to data interpretation. JA wrote the first draft of this paper and all authors were responsible for subsequent critical revision of the manuscript. D.M. is the corresponding author for this paper. This study was funded by the NIHR via the National School for Primary Care Research. J.A. was supported via funding from the Intercalated BMedSc Research Funding provided by the College of Medical and Dental Sciences, University of Birmingham and also from the Primary Care Research Trust. North Staffordshire Research Ethics Committee Ref. 07/H1204/136.
Open access
