This population- and incidence-based, matched case-control study found that the burden on primary care resources is higher in patients with AD versus their demographic- and comborbidity-comparable elderly control counterparts. The higher rates of primary care resource utilisation are evident in the AD cohort relative to GOA control patients as much as 3 years prior to AD diagnosis, with a sharp increase in the AD cohort immediately preceding the AD diagnosis that is maintained in the one year following diagnosis. This is the first study to report such a finding in the UK, and is consistent with results of other studies [13–15]. In Albert et al., the burden was measured by health care expenditures , while Ramakers et al. and Eisele et al. both used primary care consultation rates as a measure of health care burden [14, 15]. There are, however, some striking differences between the various study populations. Two of the studies [13, 14] included fewer than 100 patients with prodromal AD, while one study  had an AD cohort that was approximately half the size of that used for the current study.
The initial primary care consultation rate observed for the GOA control cohort in the current study (approximately 11 consultations per year) was consistent with general medical utilisation patterns as documented in a recent UK Department of Health report . Although this study does not reveal why patients were seeking primary care consultations, it is reasonable to assume that the increases in consultation rates and referrals for patients in the AD cohort reflect the diagnostic process. If so, diagnosis will be accompanied by an increase in health care utilisation regardless of when it occurs. Another possibility is that patients with undiagnosed AD may be less able to maintain healthy habits, such as regular medication regimens. Thus, the peak in primary care utilisation may reflect general health deterioration as the patients’ daily living skills are increasingly compromised. It is also possible that these visits may be due to sequelae subsequent to the AD, such as gait imbalances . It is notable that third-party consultations (by guardians/family/friends on behalf of the patient) as well as hospitalisations both increased, suggesting that both the diagnostic pathway and general medical decline may actively contribute to the observed increase in primary care utilisation. Further analysis is needed to elucidate the specific contribution of the AD diagnostic pathway versus medical needs.
This study also found that the AD cohort was more likely to have a diagnosis or reported symptoms of depression and/or psychosis, although the prevalence of psychosis was lower than expected in both cohorts. Recent estimates are of about 1% or higher of older-age adults exhibiting psychotic symptoms [26, 27]. The low rates of psychosis in our study and the contrasting higher antipsychotic prescriptions rates in the AD cohort versus GOA patients may be a result of prescriptions originating in a secondary care setting or prescribing to patients in the absence of a psychosis diagnosis. Our study also found lower rates of hypertension in the AD cohort than the control cohort, which is surprising since hypertension is a known risk factor for AD. In contrast, a German study reported higher rates in the AD cohort than controls . Notably, though, the proportions of patients with hypertension in the GOA and AD cohorts in our study were numerically close — statistical significance of the small difference between cohorts could be an artifact of the sample size — and may warrant further study to understand the reason(s) for the observed rates.
The initial secondary care specialty referral rates for the AD and GOA cohorts (6.9% and 5.8% in the first 6-month period, respectively; if extrapolated these would each be about 12% to 14% per year) were comparable with previously published referral rates for the GPRD (13.9% per year) . The referral rates for the AD cohort were higher than the GOA cohort throughout the study, and sharply increased in the AD cohort in the 6-month period immediately preceding diagnosis (to 17.0%); this was largely driven by an increase in psychiatric/memory clinic referrals in this time period before diagnosis.
While GP practitioners are the cornerstone of medical care, they have varying familiarity with AD and high demands on their time and attention that may limit their capacity to manage AD care. For this reason, the NHS has established memory clinics throughout the UK and recommends that AD be diagnosed within these clinics or by a psychiatric specialist, since any prescribed pharmacological therapy (e.g., AChE inhibitor) should be initiated by a specialist. Per treatment guidelines, such clinics should be the single point of referral for possible dementia diagnosis . However, this study found that about 90% of patients with an AD diagnosis in their medical record were not referred to a psychiatry/memory clinic specialist by their GP provider. It is likely that some patients were referred by non-GP physicians, such as during an unrelated hospital admission, although some of these AD diagnoses were perhaps initiated directly by the GP practitioner. As a result, there may be variability around the exact criteria used for initial diagnosis and extent of disease progression at that time.
There is growing interest in the quick, efficient diagnosis and early intervention for AD, considering the high resource utilisation prior to and post diagnosis (in the current diagnostic pathway) may be related to both the progressive nature of the disease and the often lengthy diagnosis process. Albert et al. calculated that undiagnosed AD is responsible for $130 million to nearly $200 million in excess cost to the US each year . While the number of patients with AD is much smaller in the UK than in the US, the increase in medical consultations is likely to impose a significant financial burden on the UK. There is keen interest in early treatment of neurocognitive diseases, as treatments may be most effective when structural changes to the brain are minimal ; this is also reflected in the large number of compounds under development for the treatment of early-stage AD. In addition, emergence of new clinical tools may provide the means to increase the efficiency of diagnosis, including reducing rates of false-positive and false-negative diagnoses that can contribute to the high cost and resource consumption of primary care of dementia patients .
Reassuringly, nearly 50% of the AD cohort was prescribed an AChE inhibitor within 1 year of diagnosis, adhering to the 2012 NICE treatment guidelines recommending treatment for patients with both mild and moderate to severe AD . While it is unclear at this time whether earlier diagnosis of AD would subsequently decrease the burden on the health care system, there is evidence that early AD diagnosis decreases subsequent primary health care medical costs . Our study did not address the timing of AD diagnosis as a factor in the health care resource burden on primary care. However, given the current challenges in the diagnostic pathway, combined with our evidence of the significant burden this places on primary care, it is reasonable to hypothesise that increasing the efficiency of diagnosis could improve resource and cost management for GP providers. This might be accomplished in several ways, including shortening the duration and/or streamlining the process from the time of onset of AD symptoms in the GP setting to subsequent diagnostic confirmation and mobilisation of appropriate care.
This study relied entirely on electronic medical records; eligible patients were identified from over 11 million patients in the CPRD. This population-based approach should provide a cross-section of rural/urban location, race, and socio-economic status, avoiding many biases inherent in studies focused on a specific region or clinic. There were 11,688 patients in the study population, a far larger sample size than is easily obtainable from in-person interviews. Most importantly, the resource utilisation patterns described herein occurred in the real-world context of diverse UK communities. The longitudinal design was an advantage, as it followed patients for the 3-year period prior to AD diagnosis and 1-year post. This is important in the context of typical delays that occur between the onset of AD symptoms and obtaining an AD diagnosis [6, 7].
The results of this study should be interpreted in the context of its limitations. First, while electronic medical records data are extremely valuable for efficient and effective examination of health care resource utilisation, there exist inherent limitations. Data are collected from routine practice; thus, some data may be missing and coding errors might have occurred. Second, the study analysed the CPRD database of electronic medical records for GP providers. Resource utilisation that occurred at the specialist level may not have been captured in the system, as the database does not link patient records to all specialist secondary care information (except HES). The lack of complete secondary care information limited our ability to draw conclusions about the role of the AD diagnostic pathway (versus disease progression) in resource utilisation. Future studies should include evaluation of resource utilisation at the specialist level. Third, we were also limited by the study design; inclusion in the study required a documented AD diagnosis, so patients who did not visit a physician or did not receive a diagnosis were not included in the study population. Lastly, this study matched the AD cohort with a general elderly control population by demographics and comorbidity index and adjusted for potential available confounders, but an unmeasured confounder may exist.
This study clearly shows that in the UK, AD imposes a burden on the primary care system, particularly around the time of diagnosis. Given the important role of GPs in the lives of their patients, this finding is not surprising, and is unlikely to change in the future. Because of their long-term relationships with patients, GPs are uniquely placed to observe cognitive changes in their patients over time. However, it is important to recognise the additional stress that this places on the primary care system.