Page last updated: 25-MAR-2008

General Practice - Practice Team Information (PTI)

Note of Revisions

Note of Revisions March 2008

Practice Team Information (PTI) provides national estimates of primary care workload using data from a sample of Scottish general medical practices. During 2006 a substantial review of the way in which PTI data are processed and presented was undertaken, resulting in some fairly major changes in methodology. Previously, staff recording data for PTI added a data 'modifier' field to indicate whether the consultation related to a new or pre-existing problem. This field was used to identify records for processing. Because of concerns about the completeness and reliability of the data modifier field a number of other criteria were developed to identify relevant records. Records had to relate to a face-to-face encounter, as determined by the 'encounter type' (i.e. excluding administrative entries or telephone contacts) between patients and identifiable members of the practice team (i.e. records of contacts with other or unknown practice staff, or administrative entries, are excluded). Additionally, known duplicate records were removed and records with certain Read codes were excluded if referring to administrative issues. The new method was then used to revise figures for 2003/04 and 2004/05 and to estimate new figures for 2005/06, and these were published on the PTI website in March 2007. The impact of these changes on PTI estimates is discussed in more detail in the Note of Revisions March 2007.

The new method has since become an integral part of PTI procedures. Through it the PTI dataset has been substantially strengthened, and with it the robustness of PTI-based contact and patient estimates has been improved. The drawback is that the lack of modifier information means that it is no longer possible to make a distinction between new (incident) and existing (prevalent) cases of specific conditions. Also, if records are excluded based on the status of the Read code (i.e. referring to an activity or morbidity, or to neither and hence being deemed an 'admin' code), correct allocation of the status of each Read code is of paramount importance.

In conjunction with the review of the PTI dataset, ISD carried out a complete review of the grouping used for Read codes (previously called Standard Morbidity Groupings (SMGs)). The new groupings were called Read code Groupings (RCGs) and offer a refinement of the morbidity groupings. Non-specific symptoms that do not necessarily define the condition were removed from the grouping for that condition, whereas specific activities, signs or symptoms indicative of a condition but previously not included, were added. Other previously ungrouped or very broadly grouped codes, including activity codes, were grouped into meaningful categories (such as respiratory symptoms, psychological symptoms, wound care). For codes cross-mapping to ICD-10 clinical codes (ICD-10 being a coding system commonly used in secondary care records), the categories relate directly to ICD-10 chapters or parts of chapters. Whilst revising the groupings, all Read codes were also given a status code to indicate if the code refers to a morbidity, an activity or neither of these (e.g. background or admin codes). More information regarding the impact of these changes is also given in the Note of Revisions March 2007.

A number of further revisions are introduced in the March 2008 release of PTI estimates. The main change for 2006/07 data compared to previous years is the absence of community nursing (district nurse & health visitor) information. Due mainly to ongoing changes in the way the service is organised, community nurses found it increasingly difficult to record their patient contacts for PTI. A number of practices stopped recording district nursing and/or health visitor data during 2006/07 and others considered leaving PTI because of the disproportionate amount of resource required to sustain this type of data collection. It was therefore decided to stop collecting district nursing and health visitor data from 1 April 2007, to safeguard the survival of PTI in a reduced form. Unfortunately, the completeness of the community nursing data that was still collected over the year 2006/07 was insufficient to allow calculation of nationally representative estimates. This means that there are now three years of data available on the 'full' practice team (2003/04 to 2005/06). For 2006/07 the figures shown are based on GP and practice nurse data only, and comparable figures are provided for the previous three financial years.

The new Read code groupings and status of individual Read codes were critically reviewed and a relatively small number of changes were made to groupings and status indicators for further fine-tuning. For some conditions this may lead to significant changes. The most notable example is dementia, because a number of Read codes that are included in the QOF business rules for dementia either did not map to ICD-10, or mapped to a more general grouping not including dementia. After thorough consultation it was decided that these codes should be included in the RCG for dementia. Also, for a number of Read codes the status was revised (e.g. from 'morbidity' to 'admin' or 'activity' and vice versa) to improve consistency. All analyses for previous years have been rerun using the updated Read code groupings and status allocations, so there will some (mainly minor) differences between figures for the same time period published in March 2008 compared to those published in March 2007.


Note of Revisions of March 2007 and impact as assessed on 2004/05 estimates

Summary

  • The PTI dataset was subject to a major review during 2006/07.
  • Data for 2003/04 and 2004/05 were revised and replaced previously published figures.
  • Data for 2005/06 were published for the first time.
  • PTI-based estimates of condition-specific numbers of contacts and patients consulting became more robust.
  • It was no longer possible to use PTI to derive estimates of incidence for specific diseases or conditions.
  • Standard groupings of clinical (Read) codes were revised to include categories for signs & symptoms and activities.
  • Following the dataset review, overall estimates of number of contacts in general practice in Scotland increased.

Introduction

Practice Team Information (PTI) provides national estimates of primary care workload using data from a sample of Scottish general medical practices. PTI is introduced more fully on our What is PTI? page.  During 2006 a substantial review of the way in which PTI data are processed and presented was undertaken.  As a result, figures for 2003/04 and 2004/05 were revised and accompanied newly published figures for 2005/06.  The PTI dataset was substantially strengthened, and with it the robustness of PTI-based estimates of contacts and patient numbers improved. However, it was no longer possible to make a distinction between new (incident) and existing (prevalent) cases of specific conditions. This text outlines the changes to data processing methodology and presentation, and summarises the impact that the changes had on PTI estimates.  More detailed information is available on request from the PTI team .

Changes to PTI data processing: the use of data modifiers

Historically, PTI data processing and analysis has been heavily reliant on the use of data 'modifiers'.  PTI modifiers are flags added by practice staff to the electronic clinical records to indicate whether the consultation was for a new or a pre-existing condition. These flags also identified the consultations that were relevant to PTI analysis. In the past year, ISD have undertaken a major review of the use of these 'modifiers' in practice data.

The catalyst for the review was the observation of declines in condition-specific contact rates and prevalence estimates from data collected since 2003/04, contrary to what would be expected on the basis of other data sources. This change appeared to coincide with preparation for and delivery of care under the new GMS contract.  A number of factors related to the new contract could potentially impact on PTI data collection. As part of their preparation for the new contract, many practices have reviewed clinical coding in general and QOF (Quality & Outcomes Framework) disease registers in particular. Changes in coding and the criteria for making diagnoses could affect consultation rates and prevalence estimates from PTI. Secondly, it was predicted that the additional data entry work required for the new contract might have affected the completeness and/or accuracy of recording of PTI modifiers. PTI modifiers are not necessarily required for other purposes and involve extra work and data entry for practice staff, which may duplicate data entry required for the new contract. As PTI analyses were restricted to records with modifiers, this could have had a significant impact on PTI-based figures.
In order to investigate these issues, ISD re-processed data submissions from participating practices for 2003/04 and 2004/05. A complete set of records (with and without modifiers) was used to examine the completeness of use of modifiers and to look at the impact of changes to the current rules used to retain records for analysis.

Examination of the data showed a significant decline in the completeness of modifier use coinciding with the introduction of the new contract.  Additionally, for some records, specific modifier codes were found to be inaccurate. In view of these findings, it was decided that modifiers could no longer be used as a primary basis for data processing or analysis.  Further work was carried out to develop a new methodology for selecting data for analysis and to quantify the effects on prevalence and contact rates.

In summary, the new methodology selects records that meet with the following criteria:

  • Records relating to a face-to-face encounter in the surgery or home (i.e. excluding administrative entries or telephone contacts).
  • Records relating only to contacts with GPs, practice nurses, district nurses and health visitors (e.g. records of contacts with other practice staff, or administrative entries, are excluded).

Additionally, further adjustments are made to deal with issues that arise from human error and/or computer system problems (affecting a small proportion of records overall).  Briefly, the adjustments are as follows:

  • Remove known duplicate records either:
    - those resulting from system requirements to record some items twice when practices were required to add 'modifiers' to data.
    - those that by their nature should not be recorded more than once a day (e.g. flu vaccination)
  • Exclude known administration Read codes.

The aim is to select just these records that directly relate to a face-to-face contact but exclude information recorded without the patient having been seen. It can sometimes be very hard to tell if a record indeed stems from a patient contact, or that it stems from an administrative procedure, or for example is an erroneous duplication of a valid record. Records are excluded only if these are very likely to be erroneous.

As a result of the review, the PTI dataset was substantially strengthened and the robustness of PTI-based contact and patient figures was improved. However, as data processing and analysis no longer took account of modifiers, it was no longer possible to make a distinction between new (incident) and existing (prevalent) cases of specific conditions. As a result, analyses could continue to provide contact and prevalence figures but could no longer include incidence estimates, the third and less used type of information produced from PTI data.

Changes to PTI data presentation: Read code groupings

Read codes are the recommended national standard coding system in general practice for recording clinical information (signs, symptoms, diagnoses or activities). Clinicians participating in PTI are asked to code the clinical information as specifically as possible. Although the number of codes that can be entered per consultation is not limited, generally no more than a handful of codes will be used. Nurses are asked to code an activity and an underlying morbidity (the reason for the activity being carried out), whereas GPs are generally not expected to enter an activity. Clinicians only record conditions that are being actively managed within an episode of care.  Conditions that a patient has, but which are not the reason for the contact, are not recorded as part of that particular contact.

PTI data containing the full range of Read codes can be analysed by combining Read codes for a particular condition into one umbrella grouping. These groupings have the benefit of permitting practitioners to record morbidity using the full richness of Read codes, while at the same time facilitating analysis using common, or standard, morbidity definitions.

In conjunction with the review of the PTI dataset, ISD have carried out a complete review of the grouping used for Read codes (previously called Standard Morbidity Groupings (SMGs)). The SMGs were originally based on chapter groupings from the ICD10 classification. The majority of Read codes (those starting with a letter) map directly to codes within the ICD10 classification. However there are a large number of other codes (mainly those starting with numbers) that have no agreed or standard mapping, and these have been the focus of the review of groupings. The new groupings are called Read code Groupings (RCGs) and have been revised to specifically address the following issues:

  • Refinement of morbidity groupings to remove non-specific symptoms that do not necessarily define the condition.
    · Division of the previous single 'symptoms and signs' category into smaller, more meaningful categories such as respiratory symptoms or psychological symptoms.
  • Grouping of activities not previously available.
  • Generally, use of categories which directly relate to ICD-10 chapters or parts of chapters.

Although it is now possible to produce PTI analyses based exclusively on Read codes that map to ICD-10 chapters, this would ignore a large quantity of codes covering a host of symptoms, signs, activities and relatively vague (preliminary) diagnoses. Practices continue to record activity codes (such as 'asthma review') that do not map to ICD10 codes or to previous Read code groupings, but that are likely to indicate a diagnosis (in this case asthma). While it would be more accurate to record the diagnosis leading to the consultation (asthma) as well as the activity (asthma review), this is often not done, and excluding such activity codes will result in a misleading exclusion of these consultations. Therefore routine analysis of PTI data includes these codes in the definition of the condition, despite these codes not linking to ICD-10.

Key impact on PTI figures for 2004/05

At the time of the 2006/07 review of PTI, data for 2003/04 and 2004/05 had already been published as National Statistics and supplied to customers in response to individual information requests. Detailed here are comparisons between 2004/05 estimates released prior to 27th March 2007 and revised estimates released on the 27th March 2007.

The comments below apply to 2004/05. In general, revisions to 2003/04 data follow a similar pattern, but the changes due to revision are of smaller magnitude. This is because the correct use of modifiers has declined in the last few years, affecting 2004/05 more than 2003/04.

1. Overall contacts
The data review resulted in an increase in the estimated number of general practice contacts for Scotland of approx 3.3% overall. This rise was particularly notable with practice nurses, whose estimated contacts rose by 10.9%. Estimated contacts rose in all other disciplines, but more modestly than with practice nurses; 0.6% for GPs, 2.3% for district nurses and 1.3% for health visitors.

2. Top 10 conditions
As expected, revisions to the read code groupings (RCGs) caused considerable changes in the 10 most common reasons for consultation when compared to those published previously. In general, the division of the 'Symptoms and signs and abnormal clinical & lab findings' category into smaller, more meaningful categories, had the expected effect. Also, groupings for specific conditions were refined to exclude non-specific symptoms, leading to an apparent decrease in contact rates for some conditions.
 
Full PTI:

  • 'Diseases of the skin and subcutaneous tissue' remained the most common group of conditions seen by clinicians in general practice, although contact rates for the group fell slightly.
  • Hypertension, diabetes, depression and acute upper respiratory infections remained in the top 10 reasons for consultation, while contact rates rose for all except depression.
  • 'Back and neck disorders' and 'infectious diseases' appeared in the top 10 for the first time.
  • Categories of 'Circulatory and respiratory symptoms and signs' and 'Digestive and abdominal symptoms and signs' were new in the top 10.
  • Anxiety, injuries, skin infections and asthma dropped out of the top 10.

GP:

  • 'Diseases of the skin and subcutaneous tissue' remained the most common group of conditions seen by GPs, although contact rates for the group fell slightly.
  • 'Back and neck disorders', 'Acute upper respiratory infections' and 'Hypertension' also remained in the top 10, and the rates increased when compared to those published previously.
  • Categories of 'Circulatory and respiratory symptoms and signs' and 'Digestive and abdominal symptoms and signs' were new in the top 10.
  • 'Soft tissue disorders' and 'Diseases of the upper respiratory tract' appeared in the top 10 for the first time.
  • 'Anxiety', 'Abdominal and pelvic pain', 'Infections of the skin' and 'Fatigue, tiredness, malaise and dizziness' dropped out of the top 10 list.

Practice nurse:

  • Hypertension remained the most common condition seen by practice nurses and the contact rate increased.
  • 'Diabetes', 'Asthma', 'Conduction disorders & cardiac arrhythmias', 'Diseases of the ear & mastoid process' and 'Ischaemic heart disease' all remained in the top 10 and contract rates increased for all of them.
  • 'Circulatory and respiratory symptoms and signs' now featured in the top 10.
  • 'Injuries' and 'Diseases of the skin' remained in the top 10 but contact rates dropped when compared to analyses published previously.

District nurse:

  • 'Diseases of the skin' remained the most common group of conditions seen by district nurses, and contact rates for the group increased slightly.
  • The subdivision of the old 'symptoms and signs' category allowed a more detailed picture of the conditions dealt with by district nurses, with 'Genitourinary symptoms and signs', 'Skin symptoms and signs' and 'Digestive/abdominal symptoms and signs' appeared in the top 10 in the revised analysis.
  • 'Diabetes', 'Injuries' and 'Persons encountering health services for other reasons' remained in the top 10 and the contact rates for these groups rose slightly in the revised analysis.
  • 'Multiple sclerosis', 'Conduction disorders & cardiac arrhythmias', 'Stroke' and 'Hypertension' all dropped out of the top 10.

Health visitor:
Care always must be taken when examining conditions seen by health visitors because most activities that health visitors carry out are routine checks for children and there is often no condition or morbidity as such to record for the contact. As a result, consultation rates for specific morbidities are lower than for other disciplines, and comprise a mixture of adults and some children. When comparing revised 2004/05 data with figures released previously, it can also be seen that:

  • 'Depression' replaced 'Diseases of the skin' as the most common condition seen by health visitors.
  • 'Skin symptoms and signs', 'Digestive and abdominal symptoms and signs', 'Psychological symptoms and signs' and 'Genitourinary symptoms and signs' newly appeared in the top 10.
  • 'Pregnancy related care' and 'Infectious diseases' now featured in the top 10.
  • 'Acute MI', 'Mental and behavioural disorders due to other psychoactive substance use', 'Asthma' and 'Ischaemic heart disease' dropped out of the top 10.

3. Top 10 activities for nurses
Analysis of the top 10 activities for nursing disciplines was previously presented by individual Read code. These activities were now more meaningfully grouped to give an accurate picture of the activities carried out by nurses in general practice.
 
Practice nurse:

  • In the revised analysis, 'Blood test/blood sample taken for testing' was the most common group of activities carried out by practice nurses. In previous analyses, 'venepuncture' was the most common activity and this code was included in the new grouping.
  • The contact rate for 'Influenza vaccination' increased significantly from 72.1 to 105.3 per 1,000 population. This is not surprising, since during the data review it was found that contacts for influenza vaccination were among those most likely to wrongly omit a modifier code.
  • 'Smoking cessation advice' and 'Activities relating to circulatory & respiratory symptoms and signs (S&S)' newly appeared in the top 10.
  • 'Smear' and 'Asthma review' no longer appeared in the top 10 activities for practice nurses.
  • In most cases individual Read codes were subsumed by new RCGs with the same name - this allowed us to group a small number of codes with the same meaning, whereas previously only the most commonly used individual Read codes would have been featured in the top 10 groups of activities. For example, 'Advice given' (Read code '679..') was included in the 'Advice/counselling (excluding smoking cessation advice & bereavement counselling)' group, which includes 124 Read codes. BP reading (Read code'662V.') was included in the group 'BP monitoring/reading', which includes 24 Read codes.

District nurse:

  • Wound care remained the most common group of activities carried out by district nurses.
  • The new RCG 'General patient care' included the code for 'bathing patient/shower' that previously featured in the top 10 activities by Read code.
  • 'Diabetes monitoring' replaced two codes that previously featured in the top 10 activities: 'regular insulin injection' and 'routine diabetic check'.

Health visitor:

  • The revised data showed that 'Child health care' is the most common activity for health visitors. This new group includes the Read code for 'Child health surveillance/child surveillance' that was previously the 2nd most common activity code recorded.
  • The new RCG 'Advice given (excl. Smoking cessation advice & bereavement counselling)' was the 3rd most common activity group carried out by health visitors.  This group includes individual codes for 'Advice given' and 'Health promotion' that featured in the top 10 previously.
  • 'Postpartum care & advice' was the 3rd most common activity grouping and includes codes for 'Infant feeding advice', 'Postnatal care' and 'Breastfeeding support' that featured separately in the top 10 previously.
  • 'Activities relating to psychological symptoms and signs', 'General diagnostic tests and assessments', 'Activities related to ears/auditory symptoms and signs' and 'Influenza vaccination' newly appeared in the top 10.

4. Specific conditions
The revisions to PTI data affected the estimated numbers of contact and patients for specific conditions in different ways. The changes were due to:
1) Changes to the methods used to select data for analysis, to include data that had previously been excluded.
2) The revision of Read code groupings, including refinement of morbidity groupings to remove non-specific symptoms that do not necessarily define the condition.

Comparisons detailed below are given for 2004/05 data, with the revised analysis compared to figures released prior to 27th March 2007.

Increases in contact rates and rates of patients consulting due to new data selection methodology (2004/05 revised figures compared with original 2004/05 figures)

As expected, the change to the data selection to include more relevant data resulted in an increase in the contact and patient rates for some conditions. For these conditions, the revision of the Read code groupings (RCGs) had very little effect. Groups affected included QOF conditions (angina is included in the QOF CHD indicators). This supports the view that system problems forcing double-recording of these conditions contributed to the decrease in modifier use. Specifically:

  • Angina. All ages patient rates increased by 43% for males and 40% for females. Contact rates increased by 36% for males and 34% for females.
  • Asthma. All ages patient rates increased by 19% for males and 18% for females. Contact rates increased by 40% for males and 37% for females.
  • CHD. All ages patient rates increased by 8% for males and 10% for females. Contact rates increased by 16% for males and 19% for females.
  • COPD. All ages patient rates increased by 9% for males and 7% for females. Contact rates increased by 13% for both genders.
  • Diabetes. There was an increase of 2% in all ages patient rates for both genders. Contact rates increased by 13% for males and 12% for females.
  • Epilepsy. There was a considerable increase in contact and patient rates. All ages patient rates increased by 35% for males and 44% for females. Contact rates increased by 42% for males and 47% for females.
  • Hypertension. There was a slight increase of 3% in all ages patient rates for both genders. Contact rates increased by 5% for males and females.
  • Hypothyroidism.  All ages patient rates increased by 8% for males and 6% for females. Contact rates increased by 7% for males and 5% for females.
  • Stroke and TIAs. All ages patient rates increased by 10% for both males and females. Contact rates increased by 7% for both genders.
  • Osteoarthrosis. This is not a QOF condition but the effect of the revisions follows a similar pattern to those detailed above. Patient and contact rates increased by around 2% for both males and females.

Note: While there was a marked overall increase in patient and contact rates for CHD, COPD, epilepsy, diabetes and osteoarthrosis, the revision of RCGs actually caused a small decrease in apparent rates for these conditions. This is because the revised groupings exclude some non-specific codes. Read codes for symptoms that may indicate a problem with the condition but are not recorded as a definite diagnosis at the time of contact are now included in a separate categories for symptoms and signs. For example the Read code for 'grief reaction' is no longer included in depression, but included in 'Psychological symptoms and signs'.

Decreases in patient and contact rates largely due to RCG revision

The contact and patient rates for some conditions decreased, with the majority of the decrease due to the revision of the Read code groupings (RCGs). This is because the Read codes selected for the revised grouping now exclude non-specific codes. Read codes for symptoms that may indicate a problem with the condition but are not recorded as a definite diagnosis at the time of contact are now included in a separate categories for symptoms and signs. The conditions affected in this way were:

  • Anxiety and related conditions - patient rates decreased by 25% for males, 22% for females. For example, codes for bereavement and insomnia are now included in the 'Psychological symptoms and signs' category rather than in this group.
  • Depression - there was a marked decrease in contact and prevalence rates for males (9% and 11% decrease respectively) and a decrease of 9% for both contact and prevalence rates for. For example codes for 'suicidal thoughts' and 'grief reaction' are no longer included in this group but are instead included in the 'Psychological symptoms and signs' category.
  • Dementia. Contact and patient rates for dementia decreased considerably due to the use of a new Read code grouping for this condition. This is because two codes previously included in the grouping were not specific to dementia:
    'E00..' - 'Senile and presenile organic psychotic conditions' and
    'E00z.' - 'Senile or presenile psychoses NOS'.
    With the introduction of dementia into the list of QOF conditions from 1 April 2006, it is hoped that coding will be more specific and as a result, more accurate measure of prevalence for the condition will be possible in future publications of PTI data.

Queries?
To prevent confusion, 'old' data for 2003/04 and 2004/05 were removed from the website and replaced by the revised data. The notes given here are provided to give an overview of the changes brought about by the PTI data review. We are happy to provide copies of the old data on request and to discuss the review in further detail. Additionally, if you have been provided with PTI analysis previously for the 2003/04 and 2004/05 periods, we are happy to carry out a revised analysis for you - simply contact the PTI team .

back to top button