10 PRACTICAL DIABETES Vol. 38 No. 1 Copyright © 2021 John Wiley & Sons A 84-year-old woman with type 2 diabetes for 13 years became generally unwell; long-term control of her diabetes, hypertension and dyslipidaemia was poor. Investigations revealed a chest infection and her creatinine had risen from 111 to 429μmol/L over seven days (Table 1). Plasma osmolality, and urine ketone levels were normal. She had taken a stable dose of perindopril for several years and was not taking any other nephrotoxic drugs. She had not been obviously hypotensive or hypovolaemic; renal ultrasound was normal. She had a previous history of breast cancer and deep vein thrombosis. She had recently been dyspnoeic with clear chest X-ray but raised D-dimer (1241ng/ml); computed tomography pulmonary angiogram (CTPA) was performed three days before admission with prior eGFR 46ml/min/1.73m2 and creatinine level 111μmol/L. She fulfilled the criteria for contrast induced acute kidney injury (CI-AKI), i.e. an increase in serum creatinine of 25% or 26.5μmol/L, a temporal relationship to the contrast agent (generally two to three days), and other causes of renal impairment excluded.1,2 Risk factors for CI-AKI2,3 are: age; female gender; diabetes mellitus; hypertension; renal impairment; heart failure; myeloma; albuminuria; anaemia; hypovolaemia; nephrotoxic drugs; hypoalbuminaemia; higher glucose level; and higher LDL level. The patient had several risk factors: age, gender, diabetes, hypertension, renal impairment, hyperglycaemia, raised LDL level, and ACE inhibitor use. Characteristics of the contrast (intra-arterial, non-ionic, hyperosmolar or larger volumes) also increase the risk of CI-AKI,1–3 but modern weight-based contrast dosing minimises this. Some risk factors can be modified, some cannot. Lower baseline renal function increases the risk of CI-AKI. An eGFR of 45ml/min/1.73m2 was often taken as a cut off for using intravenous contrast.1 In elderly patients, the Cockcroft Gault formula may be closer to the glomerular filtration rate than the formulae used by the laboratories4 but this patient’s baseline Cockcroft Gault eGFR was identical: 46ml/ min/1.73m2. Different guidelines give different eGFR thresholds from 30 to 60ml/min/1.73m2 for developing CI-AKI.5 Risk stratification models, often derived from cardiology patients with intra-arterial contrast and cardiac disease (both CI-AKI risk factors), may be inappropriate for other patients.6 To avoid CI-AKI, one can consider other imaging modalities, hold potentially nephrotoxic drugs 24 hours before contrast, preload with intravenous normal saline and use safest contrast.2,3 Systematic community control of glucose, cholesterol and blood pressure might have helped this patient; there was no time to achieve this as an inpatient since pulmonary emboli on CTPA can disappear within 48 hours. Treatment is avoidance, general support, and hydration; dialysis removes contrast,7 but is rarely required. The management of metformin with contrast is controversial. The Royal College of Radiologists3 cites a lack of valid evidence regarding metformin-associated lactic acidosis due to contrast, but suggests that if the eGFR is under 60, the clinician and radiologist should liaise regarding temporary metformin cessation. Other guidance8 suggests holding the metformin if eGFR is under 30 and restarting it if eGFR is unchanged, which contrasts to the metformin Summary of Product Characteristics9 that advises avoiding metformin at this eGFR and halting metformin in conditions that might alter renal function. In conclusion, contrast enhanced scans provide much useful information, but one must be aware of the hazards and their risk factors; liaison with the radiology department is invaluable.

The patient was a 75-year-old bilateral amputee with insulin-treated type 2 diabetes, hypertension and dyslipidaemia controlled on treatment, and peripheral vascular disease treated with aspirin. He was admitted having fallen while transferring from wheelchair to bed; this was a simple fall and he had normal glucose and blood pressure levels. He was an ex-smoker, but his chest X-ray was normal. He was very rapidly back to his pre-admission state and abilities. On the ward he was mobile unaided in his wheelchair, administering his own insulin; he enjoyed zipping along the smooth hospital corridors and going to the local pub either with permission and a member of staff, or with neither, but only ever drank 1 or 2 pints. A seminal paper on outcomes post amputation stated that only 5% of amputees rehabilitated well, to become independent of their wheelchair, and that more consideration should be given to surgery that optimised wheelchair rehabilitation;1 our patient was an excellent example of someone who was very able in a wheelchair. He was a bit of a rascal who spoke his mind; thus, he remained in hospital for four weeks with ongoing negotiations about returning home to his very sheltered residence who had a ‘can't do’ mentality. During his hospital stay, he developed a weak left hand (see Figure 1); stroke was diagnosed and CT scan brain showed an old small infarct in the cerebellum, but was otherwise normal. On careful examination he had a wrist drop with weak wrist and finger extension (0/5 MRC power scale) but wrist and finger flexion, and finger abduction and adduction were normal; there was numbness on the dorso-radial aspect of the hand. When examining a hand with wrist drop, it is important to support the hand in its neutral position to allow the intrinsic hand muscles (innervated by ulnar and median nerves) to not work at a disadvantage, otherwise these muscles may appear weak. It is well recognised that several entrapment neuropathies are associated with diabetes such as median, ulnar, and lateral popliteal nerves, and lateral cutaneous nerve of thigh;2 however, data on an association between radial nerve palsy and diabetes are scarce, despite the evidence that the nerves of people with diabetes are more prone to compression neuropathies.3 A leading textbook states that radial nerve palsy is not related to diabetes, but a pressure palsy had been seen following hypoglycaemic unconsciousness.4 However, Stamboulis and colleagues reviewed patients with focal neuropathies referred for nerve conduction studies; the prevalence of known diabetes in people with radial nerve palsies was 28% versus a local population prevalence of known diabetes of 7%.5 Causes of radial nerve palsy include fractured neck of humerus, gunshot wounds, and lying on the nerve due to decreased consciousness;6 none of these applied to our patient who never returned from the pub drunk, and whose blood glucose monitoring was consistently in the 5–10mmol/L range. In most wheelchair users, the weight of their two legs prevents the wheelchair from tipping over backwards. But this is a problem for double amputees; hence for double amputees, including our patient, the wheelchair rear wheels are moved further back7 to regain balance. It was noticed that as the patient was zooming around the hospital, he would move his arms far back for rapid self-propulsion and catch the inner upper arm on the wheelchair handles; this seemed the likely cause of his neuropraxia. Standard treatment of compression radial nerve palsy is to splint the wrist to allow the hand muscles to function with a good grip,6 and wait and see; this was done as well as advice to slow down. After one week, the wrist drop was markedly improved; radial nerve compression palsies are generally neuropraxic and resolve within one week. Unfortunately, one month after discharge the patient died suddenly, presumably from cardiovascular causes, exemplifying the high mortality rate of amputees.1 Accurate clinical skills and knowledge are important to avoid unnecessary consultations and investigations, particularly in these times of coronavirus pandemic. Alternatively, if one practises using pattern recognition, this just looks like a radial wrist drop.

A 65-year-old man had had type 1 diabetes for six years, with meticulous attention to self-care, frequent blood glucose testing, careful carbohydrate counting and a basal bolus regimen of degludec and Novorapid. (Figure 1.) He would generally have one mild hypo which he could self-manage per fortnight, but in clinic pointed out that he had had just five hypos in two days. He wondered about the tramadol as the cause: he had started on tramadol to treat neck pain two days before the hypos. We replaced the tramadol with oramorph, and the hypos resolved. A nested case-control analysis within the UK Clinical Practice Research Datalink created a cohort of all patients newly-treated with tramadol or codeine for non-cancer pain between 1998 and 2012, and observed them for admissions for hypoglycaemia.1 Among 334 034 subjects, 1105 were hospitalised for hypoglycaemia; compared with codeine, tramadol was associated with a 50% increased risk of hypoglycaemia, particularly during the first 30 days of use. The frequency of hypoglycaemia was 7 per 10 000 users, but these were hospitalisations for hypoglycaemia, and it is likely that less severe hypos are not being recognised.2 The data also showed, not surprisingly, that the subjects with admission for hypoglycaemia were on average elderly, with multiple comorbidities and increased concomitant drug use. A French pharmacovigilance study3 compared reports of hypoglycaemia associated with tramadol, dextropropoxyphene and codeine; it was found that there were 90 reports of hypos with dextropropoxyphene, 53 with tramadol and two with codeine -- although the population at risk is not stated, it confirms the finding that hypos are more common with tramadol than codeine; also, some subjects became hypogly-caemic without being on antidiabetic agents, and hypos generally occurred during the first few days of tramadol use. It has been noted that tramadol directly reduces hepatic gluconeogenesis and enhances peripheral glucose utilisation in diabetic rats.2 Tramadol is a commonly used analgesic; it is both a mu-opioid receptor agonist and a re-uptake inhibitor of serotonin and norepinephrine. The opioid activity is due to both the parent compound and the active metabolite which is metabolised by CYP 2D6 and gives inter-individual differences in pharmacokinetics and clinical effects. Tramadol increases risk of convulsions, and should not be mixed with other opiates, psychiatric medication or drugs that raise serotonin levels.2, 4 In the linked editorial,2 Nelson and Juurlink wrote: ‘The increased prescribing of tramadol most likely reflects aggressive marketing coupled with the perception that it is a safe analgesic not prone to abuse. Whereas the drug's analgesic effects are at best moderate, its toxic effects are dangerous and merit respect, particularly when doses are escalated.’ A useful review of other drugs that are known to cause hypoglycaemia5 includes alcohol, and pentamidine with a high risk of hypoglycaemia which is possible in the absence of diabetic hypoglycaemia treatment, and other agents with a lesser risk of hypos such as chloroquine, ACE inhibitors, beta blockers, salicyclates, and fluoroquinolones such as ciprofloxacin. Recently, warfarin has been shown to increase hypoglycaemia rates in people on sulphonylureas,6 and quinine is known to cause hypoglycaemia, particularly in the setting of renal impairment causing quinine accumulation.7 The hypoglycaemic effects of these drugs occur via various mechanisms, often in the same drug, e.g. salicylate-induced hypoglycaemia may be caused by increasing insulin secretion, increasing insulin sensitivity, displacing sulphonylureas from protein-binding sites, and inhibiting renal excretion.5 Another common drug- related hypoglycaemia scenario occurs when a patient's dia-betes has been stabilised on glucocorticosteroids which are then reduced or withdrawn. I am most grateful to the patient for sending me the scan of his record book. There are no conflicts of interest. References are available online at www.practicaldiabetes.com.

I suspect that we have all seen a scenario similar to the one shown in Figure 1: the older person admitted as a ‘stroke’ when indeed the problem has been a hypo. One could consider such a scenario as hypoglycaemia unawareness among the health care professionals concerned, but there also appears to be hypoglycaemia unawareness among older people. Blood glucose monitoring in a patient admitted as ‘stroke’, with management annotated Matyka et al.1 studied a group of elderly and young non-diabetic men as their plasma glucose was reduced by an insulin infusion. In the young men, autonomic symptoms occurred at 3.6mmol/L, and neuroglycopaenic symptoms occurred at 2.8–3.0mmol/L; however, in the elderly men, sympathetic symptoms occurred at 3.0mmol/L and neuroglycopaenic symptoms occurred at 2.8–3.0mmol/L. This suggests that the older person has less time between the development of sympathetic symptoms and neuroglycopaenia than do younger subjects. Symptoms were also less pronounced in the elderly individuals. A similar recent study in middle-aged and elderly diabetic subjects2 also showed that hypoglycaemic symptoms were less marked in the older person, and that the older person was much less likely to recognise that they were hypoglycaemic. Jaap et al. examined 132 insulin treated diabetic people aged 70 years or more;3 102 had been hypoglycaemic in the prior two months with a median of six hypos in the prior 12 months. The older person was noted to have less intense symptoms, less headache afterwards, and less autonomic features than young people. The older person had three separate groups of symptoms: impairment of coordination and articulation, light-headedness and unsteadiness, and autonomic symptoms – i.e. the symptoms were often non-specific for a frail older person who might have postural unsteadiness or hypotension concomitantly. It is recognised that patients may not recall their hypoglycaemic episodes (although their family members do),4 and older folk have extremely poor knowledge about hypoglycaemia5 so that the prevalence of hypoglycaemia in older people may be under-appreciated. Risk factors for insulin or sulphonylurea induced hypoglycaemia – which include any organ impairment, choice of hypoglycaemic agent, male gender, alcohol excess – are well known,6,7 but are they appreciated by prescribers and does it matter? The question of hypoglycaemia as a cause of permanent cognitive impairment has been considered for some time. Gold et al.8 published a series of five patients with a long history of hypoglycaemia who developed significant cognitive impairment, but the direction of causality was not certain. A further case report9 carefully assessed a patient with a significant hypoglycaemic episode who was readmitted with a second severe hypo following which his abbreviated mental test score dropped and stayed decreased. The second hypoglycaemic episode was due to the fact that the patient had restarted his sulphonylurea which had been stopped – demonstrating the importance of checking for hoarding, and of good communication to ensure that the offending agent is removed from the repeat prescription computer system. A recent study has cast more light on the subject. Whitmer and colleagues examined 16 667 Kaiser Permanente diabetic Californians from 2003–200710 who were not known to have dementia or cognitive impairment at the onset of the study (mean age 65 years), and who had been observed for episodes of significant hypoglycaemia from 1980–2002. Details regarding co-morbidities, diabetes and demography were put in a multivariate model. Admission to hospital with one hypo increased the risk of future dementia by 1.26 (95% CI 1.1–1.49), and three hypos increased the risk by 1.94 (95% CI 1.42–2.64); attending the emergency department with one hypo increased the risk of future dementia by 1.42 and two hypos increased the risk by 2.36. This study has been criticised for under-ascertainment of hypoglycaemia, and a lack of consideration of several other possible confounding factors.11 The authors accepted that one should be wary of observational data;12 however, they pointed out that the relationship between hypoglycaemia and dementia was strong, was graded and did include many confounding variables, so that their study ‘justifies caution against over-treatment of older patients with diabetes’. Analogue long-acting insulins have helped the older person considerably, but are often given by a district nurse in the late morning, several hours after breakfast, with blood glucose monitoring performed at this time. The pattern of relatively low fasting plasma glucose levels seen on glucose tolerance testing of the frail older person13 appears, in my experience, to be mirrored in clinical practice with attempts to drive down a high blood glucose with long-acting insulin resulting in fasting hypoglycaemia. Hypoglycaemia treatment is often modified, based on the HbA1c. Several studies have shown that age itself increases the HbA1c,14-16 and it has also been noted that iron deficiency itself raises the HbA1c;17 both these conditions are common in elderly patients. Thus, the clinical information available on glucose levels in older people may mislead the prescriber to inappropriately increase hypoglycaemic medication. What is the evidence for driving down glucose levels in older people? There are many observational studies showing better quality of life, better cognition, better muscle strength and fewer infections with better glycaemic control; in addition, there are several smaller studies showing improved quality of life and cognition, although Testa et al. showed benefit in a randomised controlled trial of 569 subjects with mean age 58.4 years.18 Looking at subjects entering large glycaemia trials, the mean age was 54 in UKPDS, 60.4 years in the Veterans Study,19 and 66 years in ADVANCE. The frail older person just does not get into the trials of hypoglycaemic agents. I have been involved in several trials of hypoglycaemic medication in elderly patients (Diabetes In the Very Elderly Trial, 3L studies) which were abandoned because of the difficulties with recruitment (and I do not think that it was me). Thus, hypoglycaemia in elderly patients may have serious consequences. The risks and benefits of glucose lowering in the older person are not totally clear and further research is needed; glucose lowering therapy needs to be applied in a careful, individualised approach to the older person, bearing in mind the problems of assessing glycaemia. All patients need an individualised approach, but elderly patients need it more than others. There are no conflicts of interest.

An 87-year-old man with type 2 diabetes for 12 years and insulin therapy for six years presented with left-sided facial weakness of 24 hours' duration (Figure 1). He complained of severe pain in the left ear for six months, which had not changed, and for which he had received several courses of oral and topical antibiotics; his external auditory canals were normal when examined one month previously. Left-sided facial weakness. (Permission to reproduce this photograph has been granted by the patient) Examination revealed a seventh lower motor neurone cranial lesion as well as tragal tenderness, the hallmark of malignant otitis externa (MOE). Otoscopy revealed pus and oedema of the external ear canal. A cranial CT scan (Figure 2) demonstrated obliteration of the external auditory canal, opacification of mastoid air cells, and destruction of bony cortex. Invasion into deeper structures defines malignant otitis externa.1 CT head (bone windows) showing obliteration of the external auditory canal, opacification of mastoid air cells, and destruction of bony cortex on the left Ninety percent of subjects with MOE have diabetes mellitus; proposed mechanisms include immunocompromise, microangiopathy of the external auditory canal, and increased pH of cerumen.2 MOE presents with exquisite otalgia and otorrhea. Complications include a cranial nerve palsy in up to 43% of patients – commonly the seventh, but may include the sixth to twelfth nerves (excluding the eighth). Other complications include meningitis, brain abscess, and dural sinus thrombosis.3 An initial series in the 1960s demonstrated a mortality of up to 50%,4 justifying the term malignant otitis externa; however, this has since fallen and in the most recent case series there were no deaths.3 Diagnosis is based upon clinical findings with CT or magnetic resonance imaging confirming the spread of infection outside the auditory canals in all cases.5 Occasionally, biopsy from the external auditory canal is required to exclude squamous cell carcinoma which may have similar presentation and radiology. Pseudomonas aeruginosa (the organism in this case) has been implicated as the cause in up to 95% of cases of MOE, but other organisms implicated are Staphylococci and Aspergillus species.3 Treatment involves strict glucose control, aural washing, and intravenous antibiotic therapy, typically with fluoroquinolones, for six to eight weeks.6 Surgical treatment (partial resection of the temporal bone) is rarely performed. This patient responded to intravenous antibiotic therapy, but his facial nerve function did not improve; involvement of the facial nerve does not worsen the prognosis in MOE and recovery of facial nerve function is an unpredictable indicator of successful treatment.3,6 In a diabetic person with earache, particularly if associated with cranial nerve palsies, one should consider malignant otitis externa with a thorough clinical examination.7

This chapter contains sections titled: Introduction Definition of diabetes in the elderly Why detect diabetes? The symptoms of diabetes Glycosuria Fasting plasma glucose and the modified oral glucose tolerance test Random and postprandial plasma glucose Blood glucose meters Glycosylated haemoglobin Fructosamine Diabetes prediction calculators Known diabetes Types of diabetes Metabolic syndrome Future research Conclusions References

A 24-year-old Caucasian female with type 1 diabetes mellitus for 16 years was admitted with diabetic ketoacidosis (DKA). She had a four-day history of tender frontal headache, nausea and vomiting being treated as sinusitis, with coamoxiclav 625 mg tds. She managed her diabetes with a basal bolus regimen, but had had no routine diabetic care for two years. There was one previous episode of DKA following dental surgery, but no other past medical history of note and in particular no history of miscarriage, additional medications or recent oral contraceptive use. There was no family history of thrombophilia. She was apyrexial and haemodynamically stable with Kussmaul breathing and Glasgow Coma Scale of 14, but otherwise examination was normal. She had DKA as evidenced by plasma glucose 17.3 mmol/L, blood pH 7.17, urinary ketones 4+ and Kussmaul breathing. She was treated with intravenous fluids, insulin sliding scale and the antibiotics were continued. During the 12 hours following admission, her headache intensified requiring regular analgesia and she developed neck stiffness. A CT scan of the head performed whilst considering lumbar puncture showed increased density in the superior sagittal sinus and right transverse sinus suggestive of cerebral venous sinus thrombosis (CVST) (Figure. 1), and MRI venography (MRV) (Figure. 2) confirmed thrombosis in the superior sagittal, right transverse and sigmoid venous sinuses. The maxillary and frontal air sinuses were clear on all these scans. CT scan suggestive of cerebral venous sinus thrombosis. MRI venography with absent right sinuses. CVST is an uncommon condition; estimated annual incidence is three to four per million, with 75% of cases being adult females,1 and can have a delayed diagnosis. The main risk factors include pregnancy, the puerperium, and oral contraceptive use. The risk is also increased in coagulopathies, blood dyscrasias, dehydration, malignancies, local and generalised infection, dural puncture, trauma, steroids and Ecstasy use.2 The unifying diagnosis in our patient was venous sinus thrombosis due to a coagulopathy associated with DKA. Recently, CVST has been reported in diabetes without DKA.3 Venous sinus thrombosis can present with headache, focal neurology, altered conscious level or seizures.2 The presentation can also include cranial nerve palsies and orbital oedema; interestingly, our patient developed right 6th nerve palsy and orbital oedema four days after admission. The diagnosis requires a high index of suspicion and is confirmed by MRI and venography (CT or MRI) if not diagnosed by CT. CT scan can be normal in 20% of cases but is readily available and useful for excluding other pathologies. The management of CVST in a stable patient focuses on full anticoagulation with heparin and then warfarin with a target INR of 2–3 (unless comorbidities preclude its use).4 The anticoagulation should continue for a period of six months following a first episode of CVST.1 Thrombolysis is limited to patients with a poor prognosis in centres with experienced interventional neuroradiologists and is currently limited to case reports.5 There is little evidence in the management of CVST and much of the management is based on consensus. Interestingly, the CT and MRI, tests for sinusitis with high sensitivity, showed no evidence of sinusitis; although the patient had a significant neutrophilia, no source of infection was ever found. She made a good recovery with minor residual 6th nerve palsy. She was warfarinised for six months with neurology, ophthal-mology and diabetes follow up. Repeat MRV showed almost complete recanali-sation of the venous sinuses after four months. Importantly, she was advised to register with a GP and have the support of the diabetic specialist nurse. A full thrombophilia screen will be conducted one month after she finishes her warfarin. This case highlights the need to consider CVST in diabetic patients who present with unexplained headache, focal neurology, seizures or an altered conscious level.

Ever since the 1979 National Diabetes Data Group and 1980 World Health Organization (WHO) criteria, the USA and the rest of the world have used the same criteria, apart from minor differences converting mg per deciliter to millimoles per litre. However, in 2003, the American Diabetes Association (ADA) decreased the normal fasting plasma glucose (FPG) to under 5.6mmol/L,1 whilst the WHO and the International Diabetes Federation adhere to under 6.1mmol/L.2 Does the greater range of impaired fasting glucose (IFG) on the ADA criteria discover more IFG subjects who receive a glucose tolerance test (GTT) revealing previously undiagnosed diabetes? We were particularly interested in the effect of this difference in older people, following the report that care home residents with undiagnosed diabetes often have glucose levels in the ‘normal’ range.3 Data from routine GTTs (excluding those done during pregnancy) performed within United Bristol Healthcare NHS Trust have previously been reported in Practical Diabetes International; 4 these were re-examined regarding different FPG cut-offs in middle-aged and elderly subjects (the results are shown in Table 1). ‘Elderly’ was taken as 60 years or older, following WHO criteria.5 A total of 121 younger people had diabetes (FPG ≥ 7.0mmol/L or post-challenge glucose ≥ 11.1mmol/L); of these, one (1%) had FPG <5.6mmol/L and two (2%) had FPG 5.6–6.0mmol/L inclusive. In all, 151 older people had diabetes; of these, 10 (7%) had FPG <5.6mmol/L and 11 (7%) had FPG 5.6–6.0mmol/L inclusive. In young people, changing the normal plasma glucose value has little effect since only 2.5% (95% CI 0.5–7.0%) with undiagnosed diabetes have an FPG <6.1mmol/L. On the other hand, older people with undiagnosed diabetes often have a low FPG with 10 (7%; 95% CI 3–12%) having FPG <5.6mmol/L, and 21 (14%; 95% CI 9–21%) having FPG <6.1mmol/L— i.e. 7% (95% CI 4–13%) of undiagnosed elderly diabetic subjects had FPG 5.6–6.0mmol/L inclusive. These findings concur with the DECODE group's findings that, amongst Europeans with undiagnosed diabetes, the younger fatter person was more likely to have an FPG >7.0mmol/L than an older slimmer subject.6 Importantly, the DECODE group has also shown that isolated post-challenge hyperglycaemia in the elderly person is associated with increased mortality.7 Thus in young people, the FPG is reasonable at excluding diabetes whatever normal FPG one uses. However, in older people, the lower ADA cut-off of under 5.6mmol/L as normal is needed to detect many of the diabetic subjects, but even this will still miss some subjects with undiagnosed diabetes. Unfortunately, in the 151 diabetic elderly subjects, 29 (6%: 95% CI 13–25%) did not have a diabetic two-hour post-challenge value. In older people, both fasting and post-challenge glucose values are required to comprehensively assess glucose tolerance status. Sam Mostafa Senior House Officer in Medicine*, Simon Croxson MD Consultant Physician , * Leicester General Hospital, Leicester LE5 4PW, UK, United Bristol Healthcare NHS Trust, Bristol General Hospital, Bristol BS1 6SY, UK

A 78-year-old man with type 2 diabetes for 16 years, treated with diet, metformin and gliclazide, was found to have neuropathy as evidenced by monofilament scores 0/10 and pinprick decreased mid-calf down, bilaterally. Investigations revealed vitamin B12 deficiency (see Figure). In view of the lack of any other gastrointestinal problems and the benefit of metformin therapy, replacement therapy was commenced with parenteral hydroxycobalamin without further investigation, and metformin was continued. Haematology results showing the patient's vitamin B12 level Two years later, his neuropathy was unchanged. B12 deficiency is common in elderly folk with up to 20% having impaired absorption.1 However, metformin use is associated with B12 deficiency which resolves on stopping the metformin in 50% of cases.2 Metformin-related B12 deficiency is thought to occur due to (a) inhibition of intrinsic factor secretion, (b) disruption of the ileal B12 absorption, and (c) impaired calcium availability due to metformin activity interfering with the calcium-dependent process of B12 absorption.3, 4 Diabetes is associated with peripheral neuropathy, diabetic pseudo-taboparesis (upgoing plantars and absent ankle reflexes), and cognitive impairment (probably predominantly due to cerebrovascular disease). However, B12 deficiency is very common in elderly people,1 its prevalence is significantly increased by the duration and dosage of metformin therapy,4 and it can give a similar picture. It is important to screen for B12 deficiency which may be present without anaemia or macrocytosis, and other causes of neuropathy such as alcohol and thyroid status in older diabetic people with neuropathy.5 Practical Diabetes International invites you to submit your favourite slide with clinical details for possible publication in this series.

The liver is rarely a focus of attention in textbooks or journals covering practical management of diabetes, but there are several important liver related issues for the diabetes care team, which include the use of drugs in the context of hepatic impairment, and the management of non-alcoholic fatty liver disease (NAFLD), how to diagnose it non-invasively and how to treat it. There have been significant advances in our knowledge of fatty liver which is now recognised to progress from simple steatosis, to hepatitis (non-alcoholic steatohepatitis, NASH) and ultimately to cryptogenic cirrhosis. In this issue of Practical Diabetes International, we have three expert articles to guide us on these topics along with other invaluable liver related contributions. The case report by Goenka et al. rings true. How often have we reassured our patients that ‘it's just a bit of fatty liver’? And we have been surprised to find type 2 diabetic patients with established cirrhosis on a CT abdominal scan with liver function tests (LFTs) remarkable by their normality. So although we routinely monitor LFTs as part of the regular diabetes review (usually looking at drug safety), the abnormalities can be so mild that one would probably just monitor them and ascribe these changes to the ubiquitous bit of fatty liver. Again, it is a reminder from our days at medical school that many of the conventional liver enzymes that we measure are not particularly good markers of liver function. Both Goenke et al.' s report and Preiss and Sattar's review raise the highly important question of whether to do a liver biopsy once one has excluded other causes of liver disease; this potentially dangerous procedure may well not alter management, but at present it is the only method to differentiate a simple fatty liver from an inflamed liver. A non-invasive method with high specificity and sensitivity would be invaluable. Preiss and Sattar's review documents differences in commonly used serum markers between NAFLD and alcoholic liver disease, and, along with the report by Goenke et al., documents other markers that might be used in the future to diagnose NAFLD non-invasively and to differentiate simple steatosis from NASH. Treatment of NAFLD is addressed in all three articles, but particularly by Ahmed and Byrne who also discuss the use of other drugs in the setting of liver dysfunction. Evidence suggests that pioglitazone improves features of NAFLD, although there is no licence for this. The benefit from metformin was equivocal, but Ahmed and Byrne point out that metformin is not metabolised by the liver, is not hepatotoxic and is therefore safe in liver disease; this is not recognised by the manufacturers and regulatory authorities who list hepatic impairment as a contraindication to metformin use,1 and this is certainly an option that we should be considering with our patients. The use of statins in NASH is also reviewed by Ahmed and Byrne; it appears that statin use does not worsen LFTs more than one would expect in individuals with normal baseline liver function, and may indeed improve LFTs in some. Although the prevalence of NAFLD is unknown, it is estimated to be extremely common in diabetic USA populations— e.g. up to 50%, as these three articles point out— and, with the increasing rates of obesity and diabetes in developed countries, can only get more prevalent in the population. We hope that this collection of articles in Practical Diabetes International highlights the importance of liver disease in diabetes and will aid the understanding and management of this spectrum of liver conditions.

A 53-year-old female with type 1 diabetes mellitus for 48 years was admitted with diabetic ketoacidosis (DKA) due to pneumonia. She had had poor diabetic control with multiple previous hospital admissions for hyperglycaemia or DKA and an HbA1c of 8.6%. She had treated, controlled dyslipidaemia and hypertension, a femoral-popliteal arterial bypass graft, gastroparesis and continued to smoke cigarettes. She also had a raised creatinine (160µmol/L) on admission that worsened in spite of treatment for DKA. The patient deteriorated further over the next few days requiring transfer to intensive care (ICU) despite treatment of DKA and pneumonia as per protocol. After 18 days, she complained of severe abdominal pain but further details were sketchy since she was on respiratory support. There was diffuse abdominal tenderness on examination. Abdominal ultrasound showed generalised hepatic hyperechogenicity sparing the caudate lobe, which was thought to be a fatty liver. Liver function tests were abnormal with AST 1530iu/L (normal range 6–36), alkaline phosphatase 723iu/L (NR 20–120) and bilirubin 9mmol/L (NR <17). CT abdomen (Figure 1) revealed infarcted liver. Liver infarct (arrowed) shown on CT scan of the patient's abdomen Despite aggressive treatment of her multi-organ failure including renal failure, adult respiratory distress syndrome, and an ischaemic leg treated by amputation, the patient died 20 days after admission. Hepatic infarction due to arterial insufficiency (to be distinguished from the Zahn infarction of pure portal venous obstruction and the Budd Chiari syndrome of hepatic vein thrombosis) is extremely rare due to the dual blood supply of the liver with both the hepatic artery and portal vein supply. It is thought that impairment of both these vessels is needed to cause hepatic infarction. In a series of 5420 consecutive autopsies, there were only 20 liver infarcts, and 85% of these had had both portal venous thrombosis plus systemic hypotension to reduce the hepatic arterial flow;1 our subject had needed inotropes since admission to ICU but, on the day preceding the abdominal pain, she had become hypotensive requiring an increased dose of inotropes. In the autopsy series, only 10% of subjects had combined thrombosis of hepatic artery and portal vein, and these were both iatrogenic. Most described cases of hepatic infarct were either due to trauma (accidental or surgical),2-5 or associated with atherothrombotic states such as sepsis, pregnancy, intrinsic clotting abnormalities or atherosclerosis. The first case of hepatic infarction associated with DKA, diagnosed with angiogram, isotope liver scan, and biopsy, was apparently reported in 19776 and 1978.7 In 1990, further cases of hepatic infarction were reported in diabetic patients.1, 8, 9Although varied, presenting symptoms largely included abdominal pain and elevated serum transaminases. The confirmation of hepatic infarction has been made by hepatectomy, biopsy, or clinical course compatible with infarction with angiographic or surgical evidence of hepatic arterial abnormality.10 However, a recent series has suggested that CT is the modality of choice for imaging of hepatic infarction, providing useful information regarding site, morphology, and extent of the lesion;11 indeed, in our case, the infarct was seen on CT, but not ultrasound. In most reported cases, patients made an uneventful recovery with treatment of underlying causes or anticoagulants in cases of portal or arterial occlusions. It is not known whether prophylactic heparin or full anticoagulation would avoid the development of hepatic infarction; in our case, the patient received full anticoagulation for the first 10 days in an attempt to save the leg, but none after the amputation. Further imaging of survivors has shown no sequelae. Our patient exhibited multiple risk factors for hepatic infarction including poorly controlled diabetes mellitus, DKA, atherosclerosis, hypotension and hypercoagulable state due to severe sepsis. Although outcomes are normally good, our patient's multi-organ failure leading to the hepatic infarction was always likely to prove fatal. There is a paucity of reports of hepatic infarction in diabetes in the literature which makes us believe that it is rare, but is it so rare? Patients admitted for DKA often complain of non-specific abdominal discomfort or pain, which generally resolves with treatment of DKA; hepatic infarction rarely features in our clinical differential diagnosis, and the diagnosis is frequently reached when patients are investigated for other causes of abdominal pain. We suggest that hepatic infarction is a diagnosis that should be actively considered in diabetic patients with elevated liver enzymes presenting with abdominal pain. Early diagnosis and initiation of treatment are often rewarded with a good outcome, although no randomised controlled trials validate this. We are grateful for the expert radiological assistance of Dr Paul Davison. Practical Diabetes International invites you to submit your favourite slide with clinical details for possible publication in this series.

I enjoyed the case report by Dr Rajeswaran and colleagues on resolution of diabetes following removal of phaeochromocytoma. The report is well written, concise and thorough; the case is interesting with several messages, and a succinct overview of the topic. Most epidemiological studies only consider type 1 and type 2 diabetes, although one report shows that secondary diabetes is a disease of the elderly.1 In the elderly, secondary diabetes is generally due to chronic pancreatitis (their diarrhoea is not always due to drugs) or steroids, and sometimes their glycaemic control is disturbed by thyroid disease. However, in this case, the authors were dealing with a very ‘run of the mill’ patient with ostensibly type 2 diabetes, hypertension and associated vascular disease. One wonders how many other subjects attending our clinics have secondary diabetes; one has to be aware of this possibility. In this case, the diagnosis led to a cure, but if the phaeochromocytoma had not been found, the outcome would ultimately have been fatal. Overnight (1mg) dexamethasone suppression test. Measurement of serum potassium and plasma aldosterone to renin ratio in hypertensive patients. Measurement of 24-hour urine catecholamines and/or plasma free metanephrines. Minimally invasive surgery is now widely available – e.g. for adrenal and parathyroid operations; thus the frailer patient who would not have survived previous major surgery now has another treatment option available, with improved outcomes. Having browsed around this case report, I am far clearer of the significance and management of adrenal incidentalomas; however, finding diabetes secondary to endocrine disorders amongst our standard overweight, hypertensive, diabetic patients will still require clinical suspicion.

This 87-year-old lady (Figure 1) had previously had obstructive uropathy (exact cause uncertain) treated by long-term urinary catheter, and had recently been diagnosed with type 2 diabetes. She had mild cognitive impairment, pre-meal fingerstick plasma glucose levels of 10–15 mmol/L and irregular dietary intake; thus a hypoglycaemic agent with low risk of hypoglycaemia was required. Her serum creatinine was 137µmol/L, which is not uncommon in our practice and would be suitable for metformin in many guidelines which use a 150µmol/L cut-off. However, her creatinine clearance (Table 1) was only 16 ml/min when measured (normal range 70–150 ml/min). We are satisfied with the urine collection since the catheter was in place. An elderly diabetic person. (Permission to reproduce this photograph has been granted by the patient) Since creatinine is an end product of muscle breakdown, the frail older person and others with low muscle mass will have a serum creatinine level that, taken alone, will underestimate the degree of renal impairment. A recent metformin guideline did clearly state that the 150µmol/L cut-off should be used with caution in the elderly.1 Conversely, the serum creatinine can overestimate the degree of renal impairment. Several substances can react with the Jaffe method for creatinine assay including glucose and ketones;2 thus in diabetic ketoacidosis, the serum creatinine level may be falsely raised. One can estimate the creatinine clearance from the serum creatinine, weight, age and gender using the Cockcroft Gault formula (available at http://nephron.com/cgi-bin/CGSI default.cgi), which in this 28 kg lady gives a result of 11 ml/min. The Cockcroft Gault formula is said to be unreliable in wasting, obesity, oedema, and pregnancy;3 however, it does give a more realistic picture than the serum creatinine alone in our patients. As with many investigations, one must be aware of the limitations of a single serum creatinine value. Note: Interestingly, the original Cockcroft Gault paper4 compared the calculated creatinine clearance to the measured creatinine clearance, but did not fall into the common statistical trap of stating that the calculated clearance was a good tool since it correlated closely to the measured creatinine clearance (which it would since they are both measuring the same thing). Instead the paper showed that the calculated creatinine clearance was closer to the original measured clearance than a second creatinine clearance measurement.

This 85-year-old female with type 2 diabetes for four months was referred with a two-week history of a swollen and red right big toe (Figure 1) and a small ulcer on the tip of the toe. There was no history of trauma or previous foot problems, and no clinical evidence of neuropathy or peripheral vascular disease. However, X-rays (Figures 2 and 3) showed development of frank bone destruction around the interphalangeal joint. C-reactive protein level was 27 (normal <10) and neutrophil count normal. The toe healed and returned to normal after six weeks of co-amoxiclav (Augmentin). Sausage toe1 was recently described in 14 diabetic patients with neuropathy and osteomyelitis underlying the sausage toe; only six of these subjects had osteomyelitis shown on the initial X-ray, and a further seven were diagnosed on later X-ray or isotope bone scan. None of the patients had other features of infection such as fever or leucocytosis, but the majority (11 subjects) healed with antibiotic therapy alone. As the original authors stated, a ‘sausage toe’ should warn of the possibility of underlying osteomyelitis in a diabetic foot, allowing prompt treatment. Swollen and red big toe Initial X-ray Later X-ray

The study by Paula Whitty et al. looks at issues involved in setting up the DREAM trial (Diabetes REcall And Management system) which examined the benefit of an extended diabetes register compared to a ‘usual care’ register for a control group. Instead of the diabetes register being just a list of patients, it received input from a simplified, focused register data sheet and from the biochemistry department; this register then prompted patient attendance and clinical management. One can quickly identify with some of the problems faced, particularly in relation to IT with the incompatibility of different versions of the same software in different trusts, and worries regarding security. It took longer than anticipated, but guidelines were agreed by all health care professionals, different registers were amalgamated and staff were trained on the new system. There were some problems which will always occur—such as questions over patient consent, the initial increased workload and paperwork, and the common situation where an individual's targets differ from population targets. However, it was achieved, and some aspects went better than expected—e.g. liaison with patient representatives to produce the letters to patients. Valuable lessons were learnt in terms of drawing up local guidelines, needing to be adaptable and setting up the extended computerised register, and these lessons are transferable to many other situations. The enhanced veracity and detail of information in the extended register will be a valuable research tool, based in the real world. Most diabetic people will be cared for in general practice; keeping them well depends on routinely applying straightforward interventions to all patients. The DREAM trial has set up what appears to be an effective, practical system to ensure this; we await with interest outcomes of the trial—such as glycaemic and blood pressure control. If the system is effective, it can be replicated elsewhere. However, it is not free: like other managed networks, it demands IT and administrative staff support.