Health information Flashcards
Census definition
Snapshot enumeration survey
Most complete set of population data available
allows for- making comparisons between regions, resource allocation, analysing trends, denominator for health/population statistics.
ONS is responsible for the census in England.
Census data
Household questions: Type, owner/rented, number of rooms, type of heating etc
Number of vehicles
Household members
Individual questions
Demographics
Migration
Health
Education level
Limitations to national census
- Cost
- Under enumeration: particular groups are hard to reach e.g inner city men, HMOs, non english speakers, extremes of age, military personnel, people experiencing homeless, traveler communities.
- Timeliness
- Misreporting: self reporting of data
Alternatives to census
- Rolling census
- Population projections
- Population registers: ongoing collection of data on individuals. More up to date, and good databases allow for stat analysis. However, not a single snapshot so harder to compare regions, risk on confidentiality breaches.
Routine data
Data collected routinely e.g birth and death registries
Uses: Assess burden, compare location or time, Health service assessment, research and evaluation.
+ Readily available
+ Cheap
+ Some data points very robust
+ Large numbers
+ Data sources can be linked
+ Baseline data
- Limited to what’s been collected (not a perfect representation for your question)
- Variable quality e.g poor ethnicity coding
- Access may be limited
- Delayed publication
- May be incomplete
- Often poorly presented and analysed, difference in ways completed.
Ad hoc data
Data collected for a specific purpose.
+ Specify what you need
+ Target a subgroup
+ Responsive to emerging need
+ Quality can be managed
+ Can define collection technique
+ Greater depth of stat analysis
- Costly
- May be hard to link to routine sources
- Smaller sample
- Selection bias
- Validity/reliability may be poor
Define Demography
The study of characteristics and dynamics of human populations
Key demographic factors
Population size, Age structure, Fertility, Mortality, Survival.
Fertility rate calc
no of live offspring per 1000 per year in fertile women (15-49)
Mortality calc
No died per year x length of age band
/
total individuals in age band
Describe UK gender differences
More boys age 0-4 due to higher male birth rate.
Overall more females than males due to greater male mortality from accidents and suicide (younger) then generally longer female life expectancy.
Methods for population estimates
Cohort Component method
Special group calculations
Projections
Cohort component method
Technique for mod year population estimates
- take prev mid year estimate
- Increase age by 1 year
- Add births - ONS
- Subtract deaths - ONS
- Adjust for external migration - International passenger survey
- Adjust for internal migration - GP registration
- Quality control checks
Factors that affect projections
Birth sex ratio- 1.05 males
Mortality- extrapolate historical trends
Fertility and Migrations- hard to predict. Now incorporating probabilities of random changes (outwit trends) reported as probabilistic population forecasting (with 95% CI)
Life tables
list the probability a person will die before their next birthday based on age and sex.
Estimated by ONS
Period life tables
Calculated using age specific mortality in a given year.
Shows life expectancy if a given age in a given year had that years mortality rate applied for their whole life.
Life tables for specific years are generally period life tables.
Cohort life tables
Age specific mortality rates including projections. Showing an adjusted life expectancy based on projected mortality.
Can be used in survival analysis
Applications of life tables
Proportion of people born in different years who are still alive.
Remaining life expectancy
Probability of surviving to a set age
Life expectancy can derive DALY, HALE, PYLL
Replacement fertility
The number of children each women needs to have (on average) in order to maintain the current population.
UK approx 1.7
Generally 2.1 in developed countries
Higher in global south (3.45) due to high mortality rates.
Even if this drops the effect would not ben seen quickly depending on age profile of population, changes in age mortality rates, and childbearing postponement
Factors that affect population structure
Fertility, mortality, migration,
population time bomb
Population time bomb
Situation where a low birth rates subsequently results in reduced working age people and predominantly more older age people (dependant on health and social care).
Policy to address population growth - National
Family friendly policy such as maternity/paternity leave/pay,
tax breaks,
managed migration, increasing the state pension age.
increase affordable housing
key worker policies
neighbourhood renewal
sustainable development
Policy to address population growth - International
Restrict population growth (chinas one child policy)
Promote birth control / improve family planning services
Gender equality
Research diseases of ageing
Tackle environmental problems
Sources of routine data
Death registrations
Primary care data (CQRS)
HES data
Health surveys
Disease specific registries
Death registrations
Cause of death and some demographics
Must be registered within 5 days
Cause of death coded by ONS into ICD 10 codes
Published by ONS at national and LA level.
Uses : Analyse trends and compare areas
Health needs assessment
Calc life expectancy
+ Complete and timely
+ Accurate
+ Central coding
- Cause of death often less accurate for older patients
- Changes in ICD classifications mean comparison across years hard
- Occupational advancement
STATS19
Police road traffic accidents data
Mortality from RTA
+ details of incident
+ incidents that may not end in admission
- Not all collisions are reported to police
- morbidity rated by police may be different from health care staff
CQRS
Primary care data system (designed to calc payments)
- data for QOFs
- Locally and nationally commissioned services
- provides an indication of disease prevalence
+ Useful for conditions treated exclusively by primary care
+ Payment means data mostly complete
+ Linked to QOF score
-Dependant on GP
- Designed for payment not performance management
- QOF not comparable
- QOF voluntary
Clinical Practice research data link
Collects primary and secondary data inc GP electronic record, HES, Mortality data, prescribing data, disease registers
May be linked by NHS number
Available for research purposes
+ Linked data from multiple sources, Longitudinal data, lifestyle factors
- Data incomplete, cost
Health survey for England
6000 households annually, interview and examination for BMI, BP etc and lifestyle factors.
HES
Hospital Episode Statistics
NHS inpatient admissions data, coded using ICD 10 codes locally then sent nationally in to HES database.
Used for payments from commissioners, analysis of hospital usage, development of predictive risk models, assessment of outcomes, assessment of health nee for conditions seen in secondary care.
+ Well completed, timely, can link to mortality data.
- Accuracy depends on hospital coders, variable completeness, relates to episodes not patients, only secondary care, only NHS (not private)
Sources of international data
WHO databases, Global burden of disease study, Eurostat, INDEPTH network, Demographic health surveys.
Sources of bias in population data
Selection bias- in survey data
List size- overestimate of registered patients, may lead to underestimate of service provision.
Status inflation- occupational advancement- overestimate of socioeconomic status of deceased people on registration or surveys.
Artefacts in population data
Differences between observed population and true characteristics.
Due to changes in systems e.g
- 2001- ICD 10
- 2001- new census ethnicity categories
- 2001 - wards to SOAs
ICD
Categorisation system to allow for a common language for health related topics.
Allow for consistency collection, analysis, presentation of data to enable comparisons over time and between populations.
Analysis of population health, monitoring of disease frequency, classification of deaths and hospital records, morbidity and mortality stats.
Prescribing data
Information on volumes, costs and types of medicines prescribed and dispensed.
DDD- Daily defined dose : Assumed average dose for an adult fir the main use of the drug. Rough estimate for consumption.
Unit of measurement to compare drugs. Topical, vaccines don’t have DDDS
Cost
PUs- Prescribing units: can account for age etc
Compliance
MPR- medication possession ratio, days of supply:time interval
PDC- Proportion of days covered
Why monitor prescribing data?
Cost containment.
Monitor adherence to guidelines
Detect aberrant/ inappropriate prescribing
identification of adverse drug effects
Address local priorities
Performance payments (QOF)
Sources of prescribing data and challenges
Sources- PACT/ePACT
Challenges- GP level data, no reason behind prescription or or for whom so hard to assess. Would need to audit individual records to assess further
Define pharmacovigilance
The study of how to detect, assess, understand, and prevent adverse effects of medicines.
Processes involved in pharmacovigilance
Monitoring: the use of meds day to day looking for new patterns of adverse effects
Risk assessment
Information provision
Measuring: the intact of actions taken
Systems used in pharmacovigilance
ADR reporting
Research
Information from pharmaceutical companies
Regulation authorities data
Morbidity/mortality data
Risk minimisation strategy for medications
Include warnings
Reduce indications for use
Change legal status (E.g OTC to prescription)
Data linkage: Define, uses, limitations and examples
Linking one data source with another.
Linked via a patient identifier such as NHS number or DOB and other info.
Used to: Link HES data with M&M data, link primary and secondary care data, link health with education data.
Limitations: Difficult to link without identifiers, difficult to link historical data, inconsistent styles of data collection, Jigsaw attack- data becomes identifiable, issues with consent.
Examples- England connecting for health,
Indicators for health service provision
GP, practice nurses, community midwives, consultants, IP beds per 1000 population.
Distance to health centre/specialist centre,
Waiting times for routine GP appts/surgery/specialists.
Indicators for health service utilisation
Consultations per patient, referrals per patient, ED attendances per patient, OP appts, length of stay, length of appointment
List sizes, bed occupancy,
Prescriptions, screening uptake, vaccine uptake.
Uses of mathematical modelling
Costs and benefit modelling
Clinical and managerial decision making probabilities
communicable disease outbreaks
predict disease trends
predictive risk modelling
Advantages and limitations of modelling
+ Support decision making
+ Deal with complexity
+ Create alternative scenarios
+ Model at multiple levels of detail
+ Create short, mid and long terms plans
- Relies on data availability
- Quality of data
- Incorrect assumptions = poor model
- Poor underlying technique = poor model
Predictive risk modelling
Analysis of linked pseudonymous data to predict the relationship between predictor and outcome variables.
Outcome variable is often a triple fail - suboptimal care, costly, unpleasant for the patient e.g unplanned admission, readmission, NH care.
Pre requisites to predictive risk modelling
Sufficient volume & breath of data
Routinely available data
Linked data sources
Triple fail outcome recorded within the data
Effective (cost and benefit) intervention available
Data protection act
1998
Data must be:
Processed fairly and lawfully
Obtained and used for specified lawful purposes
Minimal data for required use, & minimal time
Processed in accordance with individual rights
Secure
Transferred only to countries that offer adequate protection.
Caldicott guardianship
1997
Every NHS institution has appointed guardian to take responsibility to protect patient identifiable information.
Also:
Develop protocols to manage data
Access on need to know basis
Access must be justified
Culture of data protection