4. Cultivos de uso ilícito: los efectos en las economías campesinas y la tenencia de la
4.2. La economía campesina y los cultivos de uso ilícito
Design
This was a record linkage study with retrospective case note review carried out between a large regional mental health care provider and the corresponding regional physical healthcare trust.
Participants
Patients included were over the age of 75 with a formal diagnosis of dementia, made in a memory clinic, admitted to the physical healthcare trust’s hospital for longer than 48 hours.
Data collection rationale
Because of the local health system design, the most accurate record of a dementia diagnosis should be stored in the electronic notes (e-notes) made by the memory service, which is part of the regional mental health trust. By cross referencing health records from the mental healthcare and physical healthcare trusts, patients known to both healthcare providers were established, and subsequently individual patients with a formal diagnosis of dementia admitted to the general hospital could be identified.
Data collection technique
An electronic list of all people over the age of 75 admitted to the physical healthcare trust during May 2014 was obtained, they were identified only by their National Health Service (NHS) number, and appeared in order of date of admission. The list was provided by the trust health informatics department. May 2014 was chosen as it was a month that typically has an average number of admissions for the year.
The list was randomly ordered using computerised random number allocation, and the NHS numbers were cross-referenced against the mental health trust’s psychiatric e-notes database. Randomising aimed to reduce any effect caused by the day of the week of admission, on types of admission, (more elective admissions on weekdays etc.). This established the number of cases admitted to the general
hospital, known to the mental health trust with any recorded mental health diagnosis (e.g. depression, schizophrenia, dementia). The e-notes system could not automatically identify those with dementia - the target population. To do this, the psychiatric e-notes of patients with a linked record were systematically hand searched. Hospital letters, patient notes and discharge summaries were scrutinised for any formal diagnosis of dementia, and the dementia subtype as recorded at patient’s assessment with the memory service were noted. Where severity was indicated it was recorded by reviewing the most recently completed standardised objective memory test score, and hospital letters. Patients who had undergone formal diagnostic assessment for dementia by a Psychiatrist and had a formal diagnosis of dementia documented in the psychiatric notes were allocated a study identification (ID) number.
To calculate how accurately the general hospital documented dementia diagnosis, and the frequency of documented psychological symptoms, somatic symptoms, delirium, distress and the use of symptom recognition tools, the general hospital notes of the patients with dementia identified above were scrutinised. To do this, the general medical notes of the people identified with dementia were obtained from the medical records library. Notes were obtained in batches of 20 to reduce clinical disruption. Each set of paper notes contained medical and nursing documentation relating to the May 2014 admission, which was hand searched.
Where notes were missing due to clinical activity (for instance the notes were in an outpatient clinic or a ward), the patient number was noted and the notes re- requested at a later date. Where notes were archived, for instance, because of patient death, they were retrieved by the medical records department.
The following inclusion criteria were applied to the medical notes review: • Medical notes were present and available for review during the data
collection period.
• The May 2014 admission nursing and medical notes were present.
Where hospital notes were unavailable or incomplete the case was noted and excluded.
Each set of paper notes related to an individual patient and contained medical and nursing documentation relating to the May 2014 admission to hospital. For each patient, the medical notes, nursing notes and electronic discharge advice notifications were hand searched. This provided a record of the admission details, all significant diagnoses, nursing observations, medical investigations and symptoms at the point of admission, and during the hospital stay.
Data collection
The following information was documented anonymously on the data collection form displayed in appendix 1.
• Demographic details including, age, reason for admission, length of admission, and medical speciality providing care.
• Evidence of documented dementia diagnosis, subtype and severity. • Documented psychiatric symptoms, somatic symptoms and delirium. • Documented evidence of distress1 (Regnard et al. 2007; Cohen-Mansfield,
1997).
• The use of any dementia specific standardised symptom recognition tools or treatment algorithms.
To avoid any ambiguity a diagnosis or symptom was only counted if the treating team had specifically written it. Commonly used euphemisms (for example, confusion) were not recorded.
Sample size
From expected dementia and symptom prevalence data, prior to data collection it was estimated that 100 sets of medical case notes (following exclusions), would be sufficient to demonstrate the prevalence of documented dementia and symptom frequency (primary outcome) with acceptable 95% confidence intervals (CI). However, it was planned that if greater numbers were necessary more from an available 200 would be reviewed. To check this, once 100 medical notes had been
1Distress was defined as any language describing a negative emotional or behavioural state, or any language describing distress or agitation as defined by the DisDAT or Cohen Mansfield Agitation Inventory.
reviewed the data set was analysed, then re-analysed after each subsequent batch of 20. Once the confidence intervals were consistent, or unlikely to demonstrate significance without an unfeasible increase in numbers, data collection ended.
Data analysis
To calculate diagnostic accuracy, the percentage of dementia diagnoses, subtypes and severity documented by the general hospital was compared with the psychiatric e-notes, proportional agreement and kappa coefficients (ƙ) were calculated to demonstrate inter-rater agreement. Associations between the speciality of the team providing care and dementia severity with the accuracy of dementia diagnosis recording were tested using chi-squared test.
Descriptive statistics were used to display the documented symptom frequency, percentage and 95% confidence intervals. Associations between the speciality providing care and the frequency of symptom documentation were tested using chi- squared test. Where expected cell counts did not meet approximation for chi- squared test, Fisher’s exact test was applied.
No direct comparisons between this data set and previously published prevalence data could be made due to differing methods and outcome measures (Sampson et al. 2014; Sampson et al. 2015; Hessler et al. 2017; Fick et al. 2002).
Descriptive statistics were used to display reported distress in the study group. To test for associations between distress reporting and symptom reporting, chi-squared test was used. Where data did not meet applicable standard approximations for chi- squared test, Fishers exact test was used.
All statistical data was generated using STATA (SE) software.
Data protection
Patient identifiable data in the form of medical and psychiatric notes were accessed only by the researcher, who held a contract with both the physical health and mental health trust. The psychiatric e-notes system was accessed using the mental health trust server via the secure general hospital server. The general hospital paper notes and electronic discharge notes were only accessed and reviewed in the hospital
medical records library. At no time was patient identifiable data removed from the hospital site.
In the initial stages of the study lists of patients were generated in the form of NHS numbers. This is a 10-digit code that identifies a patient and is universal across all NHS trusts. Cases known to both the general hospital and the mental health trust were identified and sourced using the NHS number. Once the case notes had been sourced the NHS number was no longer of use and was deleted from the study data, leaving only the study ID number to identify patients with, thus anonymising the data.
Data collection tools contained no patient identifiable data. All completed forms were stored securely in a locked cabinet on hospital property until the data were transferred to an electronic format, at which point completed data collection tools were destroyed.
All electronic data was kept in a password protected file on the password protected hospital secure server. All data collection, storage and use complied with the Data Protection Act (1998), and the University of Leeds information security policy.
Ethical considerations
During this study, there was no direct contact with participants, and there were no direct risks to individual research participants. The information accessed was part of the participants’ confidential records. The general hospital trust’s research committee confirmed that as the researcher was is a member of the direct care team of the participants, and data was intended to be used to improve patient services, identifiable patient information can be accessed without individual patient consent.