key: cord-0962372-qu5szvgq
authors: Muzambi, Rutendo; Bhaskaran, Krishnan; Smeeth, Liam; Brayne, Carol; Chaturvedi, Nish; Warren-Gash, Charlotte
title: Assessment of common infections and incident dementia using UK primary and secondary care data: a historical cohort study
date: 2021-07-03
journal: Lancet Healthy Longev
DOI: 10.1016/s2666-7568(21)00118-5
sha: 0caff8c0f69193eb731b3da98f0c09353d407420
doc_id: 962372
cord_uid: qu5szvgq

BACKGROUND: Common infections have been associated with dementia risk; however, evidence is scarce. We aimed to investigate the association between common infections and dementia in adults (≥65 years) in a UK population-based cohort study. METHODS: We did a historical cohort study of individuals who were 65 years and older with no history of dementia or cognitive impairment using the Clinical Practice Research Datalink linked to Hospital Episode Statistics between Jan 1, 2004, and Dec 31, 2018. Multivariable Cox proportional hazard regression models were used to estimate the association between time-updated previous common infections (sepsis, pneumonia, other lower respiratory tract infections, urinary tract infections, and skin and soft tissue infections) and incident dementia diagnosis. We also tested for effect modification by diabetes since it is an independent risk factor for dementia and co-occurs with infection. FINDINGS: Between Jan 1, 2004, and Dec 31, 2018, our study included 989 800 individuals (median age 68·6 years [IQR 65·0–77·0]; 537 602 [54·3%] women) of whom 402 204 (40·6%) were diagnosed with at least one infection and 56 802 (5·7%) had incident dementia during a median follow-up of 5·2 years (IQR 2·3–9·0). Dementia risk increased in those with any infection (adjusted hazard ratio [HR] 1·53 [95% CI 1·50–1·55]) compared with those without infection. HRs were highest for sepsis (HR 2·08 [1·89–2·29]) and pneumonia (HR 1·88 [1·77–1·99]) and for infections leading to hospital admission (1·99 [1·94–2·04]). HRs were also higher in individuals with diabetes compared with those without diabetes. INTERPRETATION: Common infections, particularly those resulting in hospitalisation, were associated with an increased risk of dementia persisting over the long term. Whether reducing infections lowers the risk of subsequent dementia warrants evaluation. FUNDING: Alzheimer's Society, Wellcome Trust, and the Royal Society.

cohort studies/ or longitudinal study/ or follow-up study/ or prospective study/ or retrospective study/ or cohort.ti,ab. or longitudinal.ti,ab. or prospective.ti,ab. or retrospective.ti,ab.

Case-Control Studies/ or Control Groups/ or Matched-Pair Analysis/ or ((case* adj5 control*) or (case adj3 comparison*) or control group*).ti,ab.

Incidence/ or incidence.ti,ab,kw.

28.

(hazard ratio or HR or odds ratio or relative risk or RR).ti,ab.

29. 25 or 26 or 27 or 28 30. 14 and 24 and 29

outcome exposure biasing path causal path ancestor of outcome ancestor of exposure and outcome Legend:

Infections were identified using primary care data from the Clinical Practice Research Datalink (CPRD) and secondary care data from Hospital Episode Statistics (HES). In order to confirm a diagnosis of infections and capture more serious infections, urinary tract infections (UTIs) and skin and soft tissue infections (SSTIs) were defined using a Read code and a prescription for antibiotics on the same date as a diagnosis of infection. Sepsis, pneumonia and other lower respiratory tract infections (LRTIs) were not defined using a prescription for antibiotics as these infections can also be caused by viruses. Infections identified in HES were not linked with a prescription for antibiotics as medication data is not available in HES and hospitalised infections are likely to be more serious than those diagnosed in the community.

To avoid misclassification of dementia, we excluded the first 3 months after an infection. That is, when an individual was diagnosed with an infection, they exited the study for 3 months and re-entered after the 3-month exit period.

For instances in which an individual was diagnosed with two different types of infections on the same date, a hierarchical approach was used based on the infection type and data source. If two different infections were diagnosed on the same date in CPRD and HES, the infection recorded in HES was used as the primary diagnosis. Then, if sepsis was diagnosed on the same date as another infection, sepsis was the primary diagnosis. If pneumonia and other lower respiratory tract infections were diagnosed on the same date, then the pneumonia diagnosis was taken. If two infections were diagnosed in hospital and GP records, the infection diagnosed in hospital was taken.

Individuals who were diagnosed with subsequent infections during the 3-month exit period remained out of the study 3 months from subsequent infection diagnosis.

We linked CPRD to patient-level Index of Multiple Deprivation 2015 (IMD 2015) . The IMD 2015 is a measure of relative deprivation for small areas across England based on seven domains which can be divided into groups ranking from least to most deprived.(1) Socioeconomic deprivation was defined using patient-level Index of Multiple Deprivation (IMD) which includes only English GP practices. Socioeconomic deprivation was categorised into quintiles; 1 least deprived and 5 most deprived.(1)

Ethnicity was categorised as follows: White, South Asian, Black and Mixed/Other. Using CPRD, we used the most commonly recorded ethnicity, then we used the most recent ethnicity where several ethnicities were recorded. When ethnicity in CPRD was missing, we then used ethnicity recorded in HES. The algorithm we used for ethnicity in CPRD and HES has been previously described. (2) BMI and smoking status

We derived BMI at entry into the study from the CPRD additional details file and defined using methods previously described. (3) We excluded records when an individual was under 16 years, during pregnancy or records under 20kg. BMI was calculated using weight records with height recorded on the same date. However, if a height record on the same date as weight record was not available, we used an older height record. If not available, we used a future height record. The following cut-off points by the World Health Organisation (WHO) were used to define BMI; underweight (<18.5), normal weight (18.5-24.9), overweight (25.0-29.9), obese (30.0>). Smoking status was defined in CPRD, using Read codes and data from the additional details file.

Smoking and BMI status were assigned using the nearest record in the period of -1 year to +1 month from start of follow up, if available (best). If not available, the second option was to use the nearest record in the period +1month to +1 year after start of follow up. If not available, the third option was to take the nearest record before -1 year from start of follow up and if not available the least best option was to take any nearest record after +1 year from start of follow up.

Depression and anxiety were also defined in CPRD and HES. In CPRD, we included individuals with: 1) a diagnostic or symptomatic Read code for depression or anxiety and 2) a prescription for antidepressants or medications indicated for treatment of anxiety in the British National Formulary (BNF) within 90 days of clinical code.

Depression and anxiety were defined as above due to the changes in diagnosing depression in the UK primary care given that in 2006 GPs switched from using diagnostic to symptomatic codes (4). Additionally, from 2004 antidepressants were no longer routinely prescribed for mild depression, therefore to increase the likelihood of capturing those with more severe depression, we defined depression with a clinical code and prescription for antidepressants (5) . Given that depression and anxiety have overlapping clinical codes and medication, we combined the variables for depression and anxiety together. In HES, we used ICD-10 codes alone as individuals diagnosed in hospital are more likely to be severe than those in diagnosed in primary care.

We defined diabetes using diagnoses for diabetes mellitus (type 1, type 2 and unspecified) and codes for diabetes complications. To define diabetes, we used Read codes in CPRD and ICD-10 codes in HES. Codes for gestational diabetes, secondary diabetes such as "diabetes mellitus induced by steroids" and diabetes care codes were excluded.

We defined polypharmacy as the concurrent use of 5 or more medications using BNF chapters. We excluded vaccines and devices that do not administer medication. Medication use was captured in the 12 months prior to baseline. Infections are unlikely to be causally associated with this type of dementia Primary analyses repeated defining all infections in CPRD with a diagnostic code and prescription for antibiotics

To improve the accuracy of our infection definition Primary analysis repeated excluding individuals diagnosed with infections from two different sites (e.g. skin and soft tissue infections and pneumonia) on the same date To avoid any potential biases introduced from including these infections.

We excluded codes relating to symptoms of cognitive impairment from analysis of infections and cognitive impairment anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors, systemic corticosteroids and polypharmacy. **Additionally adjusted for atrial fibrillation, myocardial infarction and stroke. ***Fully adjusted model additionally adjusted for BMI

HR, hazard ratio. Adjusted for age, sex, patient level IMD, calendar period, ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors, systemic corticosteroids and polypharmacy. 1·70 (1·61-1·79) HR, hazard ratio; Age as the underlying time scale. ** Adjusted for age, sex, ethnicity, patient level IMD and calendar period *** additionally adjusted for ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors, systemic corticosteroids and polypharmacy. Likelihood ratio test for interaction comparing models with and without interaction term between infections and diabetes, p=0.00014. 

1·34 (1·30-1·38) HR, hazard ratio; LRTIs, lower respiratory tract infections (excluding pneumonia); UTI; urinary tract infection, SSTI, skin and soft tissue infection *Age as the underlying time scale. ** Adjusted for age, sex, patient level IMD and calendar period *** additionally adjusted for ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors, systemic corticosteroids and polypharmacy. ratio; LRTIs, lower respiratory tract infections (excluding pneumonia); UTI; urinary tract infection, SSTIs, skin and soft tissue infection *Age as underlying time scale. ** Adjusted for age, sex, patient level IMD and calendar period *** additionally adjusted for ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors, systemic corticosteroids and polypharmacy. sex, patient level IMD and calendar period *** additionally adjusted for ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors, systemic corticosteroids and polypharmacy. Table 10 : Crude rate and hazard ratios for the association of common infections and dementia, with interaction between age and nonproportional covariates (sex, ethnicity, patient-level IMD, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, chronic kidney disease, hypertension, traumatic brain injury, benzodiazepines and polypharmacy) .55) *Age as underlying time scale. ** Adjusted for age, sex, patient level IMD and calendar period *** additionally adjusted for ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease , diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors, systemic corticosteroids and polypharmacy. 1.50 (1.47-1.53) HR, hazard ratio; LRTIs, lower respiratory tract infections (excluding pneumonia); UTI; urinary tract infection, SSTIs, skin and soft tissue infection *Age as underlying time scale. ** Adjusted for age, sex, patient level IMD and calendar period *** additionally adjusted for ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors, systemic corticosteroids and polypharmacy.

English indices of deprivation 2015 : Department for Communities and Local Government,; 2016 [cited Department for Communities and Local Government

Completeness and usability of ethnicity data in UK-based primary care and hospital databases

Representativeness and optimal use of body mass index (BMI) in the UK Clinical Practice Research Datalink (CPRD)

Changes in rates of recorded depression in English primary care 2003-2013: Time trend analyses of effects of the economic recession, and the GP contract quality outcomes framework (QOF)

Identification of mental health and quality of life outcomes in primary care databases in the UK: a systematic review

UTI; urinary tract infection, SSTIs, skin and soft tissue infection *Age as underlying time scale. ** Adjusted for age, sex, patient level IMD and calendar period *** additionally adjusted for ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke, traumatic brain injury, benzodiazepines, proton pump inhibitors

UTI; urinary tract infection, SSTIs, skin and soft tissue infection *Age as underlying time scale. ** Adjusted for age, sex, patient level IMD and calendar period *** additionally adjusted for ethnicity, smoking status, heavy alcohol consumption, anxiety and depression, severe mental illness, inflammatory bowel disease, multiple sclerosis, rheumatoid arthritis, psoriasis, asthma, chronic kidney disease, chronic liver disease, chronic obstructive pulmonary disease, diabetes mellitus, heart failure, hypertension, obstructive sleep apnoea, stroke