The challenge of measuring multiple conditions

One challenge of researching multiple conditions is that there is no agreement on which, or how many, conditions should be considered. Logically, the longer the list of conditions of interest, the more people will be identified as having multiple conditions, leading to differing estimates of prevalence. We used a list of 36 conditions, selected because they lead to a significant need for treatment or have been linked to poorer quality of life, poorer functioning or greater risk of premature death (Box 1). The list includes long-term and recurring conditions (eg depression, diabetes) as well as risk factors (eg hypertension) and symptoms including chronic pain (which has been referred to as a painful condition,). The conditions are all recorded by GPs, and have been commonly used in recent studies in England.,

Box 1: List of 36 conditions used in this analysis

  • Alcohol problems
  • Other psychoactive substance misuse
  • Anorexia or bulimia
  • Asthma (currently treated)
  • Atrial fibrillation
  • Blindness and low vision
  • Bronchiectasis
  • Chronic kidney disease
  • Chronic liver disease
  • Chronic obstructive pulmonary disease (COPD)
  • Constipation (currently treated)
  • Coronary heart disease
  • Dementia
  • Depression, anxiety and other neurotic, stress-related and somatoform disorders
  • Diabetes
  • Diverticulosis
  • Epilepsy
  • Hearing loss
  • Heart failure
  • Hypertension
  • Inflammatory bowel disease
  • Irritable bowel syndrome
  • Learning disability
  • Migraine
  • Multiple sclerosis
  • New diagnosis of cancer within last 5 years
  • Painful condition (on prescription-only pain medication)
  • Parkinson’s disease
  • Peripheral vascular disease
  • Prostate disorders
  • Psoriasis or eczema
  • Rheumatoid arthritis, other inflammatory polyarthropathies and systematic connective tissue disorders
  • Schizophrenia (and related non-organic psychosis) or bipolar disorder
  • Stroke and transient ischaemic attack
  • Thyroid disorders
  • Viral hepatitis

Our analysis used a pseudonymised database of electronic health records, the Clinical Practice Research Datalink. We obtained approval for this study from an independent committee (ISAC protocol reference number 17_150RMn2). Individuals cannot be identified from these records and data confidentiality was maintained by our rigorous adherence to strict data-handling procedures in a secure data environment. More information on data security is available on the Health Foundation’s website.

We used a random sample of 300,000 patients registered with a GP, a sample large enough to carry out our study robustly, and which, from previous studies, we know is broadly representative of the population of England. For each patient in the dataset we counted the number of conditions that appeared in the primary care records on 1 April 2014, using published code lists. We then calculated the use of health services between then and 31 March 2016 (a 2-year period) and estimated the cost of these services for every patient in the sample.

Using this approach, we could estimate the total use and cost of their NHS care over a range of primary and secondary care services. This included hospital admissions, visits to emergency departments, outpatient appointments, primary care consultations and primary care prescriptions. Specialised commissioning, specialist mental health care and community care are not included in these records.

Our analysis uses a large linked dataset, with a comprehensive picture of conditions from primary care records, but there are limitations with our approach. In most cases, the dataset didn’t record the severity of the conditions, and we didn’t track how long the patient had the condition, or whether they developed additional conditions during the 2-year follow-up period. How long a patient has a condition might be important. For example, a person may have more interaction, and thus hospital and GP activity, with the NHS when they are first diagnosed than later in the course of their disease (eg for cancers).

Previous Next