Effect of the UK Incentive-Based Contract on the Management of Patient
the Department of General Practice and Primary Care (C.R.S., P.C.H., K.L.), University of Aberdeen, and the Department of Clinical Pharmacology (D.W.), Grampian Universities Hospital Trust, Foresterhill, Aberdeen, Scotland.
Abstract
Background and Purpose— We wished to ascertain whether a new contract based on financial incentives for general practitioners has been associated with improved recording of quality indicators for patients with stroke and whether there was evidence of any difference in change between sex, age, and deprivation groups.
Methods— In a serial cross-sectional study, patients from 310 general practices with a computer record of transient ischemic attack or stroke in Scotland were analyzed for their recording of quality indicators before and after the introduction of a new quality-based contract on March 31, 2004. Multivariate analyses were used to explore any differences in recording between age, sex, and deprivation groups.
Results— Documentation of quality indicators increased over time, with absolute increases for individual indicators ranging from 32.3% to 52.1%. There was a large increase in the documentation of quality indicators among the oldest patients (>75 years) and the most affluent patients. This tended to attenuate age groups differences and to exacerbate differences between deprivation groups. Women tended to have larger increases in documentation than men; however, sex differences persisted, with women less likely than men to have smoking habits recorded (adjusted odds ratio, 0.87; 95% confidence interval, 0.81 to 0.95) or to receive antiplatelet or anticoagulant therapy (adjusted odds ratio, 0.93; 95% confidence interval, 0.86 to 0.99).
Conclusions— The recording and management of quality indicators among patients with stroke increased substantially. However, inequitable care exists, which may have important implications for female, older, and more deprived subgroups in terms of stroke recurrence and mortality.
Key Words: epidemiology healthcare policy risk factors stroke stroke management
Introduction
Virtually all individuals resident in Scotland (including children) are registered with primary care, which is free at the point of contact and which provides first-line and continuing posthospitalization care of patients. Access to secondary care is usually obtained through a primary care practice, and even when a patient is admitted to hospital (eg, because of an emergency), details of the hospital stay are reported back to the patient’s primary care practice. So far, studies of stroke care have focused on hospital management, with very little work done in primary care.1 We were the first to show that primary care management of patients with stroke/transient ischemic attack (TIA) may be suboptimal.2
In April 2004, a quality-based general medical services (GMS) contract was introduced to Scottish general practice, which reduced the proportion of income of general practitioners (GPs) derived from per capita payments and increased the proportion (&23%) derived from providing specific aspects of care, such as targets based on quality indicators.3 The new contract provides payments to practices to develop an accurate register of patients who have had a stroke (because a complete and accurate register is a prerequisite for monitoring patients) and for the recording of smoking habits, blood pressure, and cholesterol levels of patients on the register. Further payments are made for reaching a number of specific treatment targets, eg, blood pressure control. The recording of quality indicators is based on the presence of disease rather than the demographic (eg, age, sex, or deprivation) characteristics of the population served. Initial payments were based on the achievement of targets 1 year after the introduction of the contract.
In April 2000, the Primary Care Clinical Informatics Unit was created as part of national primary care information management initiatives, including the Scottish Clinical Information Management in Primary Care and its Scottish Programme for Improving Clinical Effectiveness (SPICE).4 The role of the Primary Care Clinical Informatics Unit is to help GPs understand their clinical information needs through a variety of feedback reports based on their practices’ extracted data. SPICE care management screens are used by clinicians to systematically record data about a number of chronic diseases, including stroke. The data include information required for the new contract.3 Diagnostic criteria are not given; instead, diagnoses are recorded through routine practice, which, for major conditions like stroke, is often after investigation and input from hospital-based specialist colleagues.
Although few evaluations of incentive systems have been conducted,5 a recent study of chronic disease care carried out in New York suggested that financial incentives to primary care physicians leads to improvements in objective quality-of-care measures.6 Substantial improvements have also been seen in the quality of care provided by general practices for asthma, coronary heart disease, and type 2 diabetes as a result of systematic quality improvement initiatives in the United Kingdom.7
We hypothesized that during the period when the new quality-based contract was being prepared and introduced, the recording of computerized information about key components of care for patients who have had a stroke would improve. As a secondary aim, we also wanted to see whether there was any evidence of a change in differences between sex, age, and deprivation groups before and after institution of the new contract.
Subjects and Methods
Anonymous retrospective data from all 310 of the 850 Scottish practices that use the General Practice Administrative Software System and that participate in SPICE were obtained in November 2005. These 310 practices were self-selected; however, they have been shown to be representative of all Scottish practices.8 The completeness and accuracy of morbidity and repeat prescribing data in General Practice Administrative Software System practices have been reported previously.9 The long-term nature of the database and its clinical focus help ensure that initially uncertain events tend to be confirmed or refuted over time, especially those requiring continued care, like stroke. From the accumulated data, we identified everyone who had a computer record of a TIA (read codes G65 to G654, G656 to G65zz) or stroke (including cerebral hemorrhagic; read codes G61 below but not G617, G66, and below) and nonhemorrhagic stroke (read codes G63y0-1, G6760, G6w, G6x, G64, and below) on March 31, 2004 (1 year before introduction of the new contract in April 2004, designated as the "precontract" period in this article; total population at risk 1 806 266 patients) and March 31, 2005 (1 year after introduction of the new contract in April 2004; designated as the "postcontract" period; 1 775 397 patients). All registered patients with a recording of stroke before the 2 time points were included in the analyses.
Key characteristics of every identified person at each time point were determined, as follows: sex, age (<64, 65 to 75, or 75+ years), number of stroke-related comorbidities (diabetes [C10 and below], hypertension (G2, G20 and below, G24 to G2z], atrial fibrillation [G573 and below], coronary heart disease [G3 to G3401, G342 to G366., G38 to G3z], heart failure [G58 and below], and peripheral vascular disease [G73 and below]; (0, 1, 2, or 3+), and deprivation status based on postal code (Carstair’s DEPCAT categorization, which uses household overcrowding; unemployment, social class, and proportion of all persons in private households with no car as indicators of poverty,10 with deprivation quintile 1 as being least deprived to 5, being most deprived). We also determined for strokes occurring in the 15 months before each time point whether the event had been confirmed by computed tomography or magnetic resonance imaging by looking for documentation of such scans. In addition, for all stroke patients, we looked for a recording before the 2 time points for stroke-related quality indicators defined in the new contract: whether the person’s computer records contained details about the stroke diagnosis; smoking habits (current, ex-smoker, or never smoked) and (where appropriate) provision of smoking cessation advice; measurement and (where appropriate) control of cholesterol (to 5mmol/L) and blood pressure (to 150 mm Hg systolic and 90 mm Hg diastolic); provision of antiplatelet or anticoagulant therapy (including over-the-counter salicylate) to patients with nonhemorrhagic stroke or TIA; and provision of influenza vaccination.3 The time periods used for our data analyses were determined by the GMS contract. When 1 entry existed for a particular item, we used the most recent entry.
We excluded from the analysis anyone with a record of "exception codes," which indicate that the person has refused or been considered unsuitable to have any quality indicators measured. Patients were also excluded from particular analyses when an exception code pertaining to a quality indicator existed, eg, for aspirin treatment, if there was a recorded contraindication or allergy to this medicine.
Statistical Analysis
Binary logistic regression was used to determine odds ratios (ORs) and 95% confidence intervals (95% CIs) for the recording of quality indicators among different sex, age, and deprivation groups, adjusted for potential confounding by sex, age, number of stroke related comorbidities, deprivation, and practice (except where a factor was itself being explored). Patients with missing data (eg, smoking status) were excluded from the analysis of that factor. Where appropriate, 2 tests were used to compare differences between groups. For clarity, all proportions are presented to 1 decimal place and ORs, to 2 decimal places. All analyses were performed with SPSS for Windows 12.0 (SPSS Inc).
Results
By March 31, 2004 (precontract), 21 901 patients had a computer record of any stroke or TIA (1.2% of everyone registered with the practices; 95% CI, 1.1% to 1.2%). Forty-six patients had a computer record of an exception code. By March 31, 2005 (postcontract), the corresponding number of patients was 32 401 (1.8%; 95% CI, 1.8% to 1.8%), of whom 2565 had a computer record of an exception code. There was a greater proportion of female, older, and less deprived patients with stroke or TIA before the contract compared with after the contract. Stroke/TIA patients were also more likely to have a greater number of comorbidities recorded after the contract (Table 1).
After the contract, men were more likely to have diabetes, peripheral vascular disease, and coronary heart disease and less likely to have hypertension than women (Table 2). The oldest and most deprived patients were more likely to have a recorded diagnosis of all of the comorbidities analyzed. One exception was heart failure, for which there was no difference between the different deprived groups in this diagnosis.
Male, the oldest, and the most deprived patients tended to have 3 or more comorbidities recorded than women (men n=2621, 16.1%; women n=2159, 13.4%), the youngest (>75 years n=2647, 17.6%; <65 years n=634, 8.0%), and least deprived patients (deprivation quintile 5, n=749, 16.0%; deprivation quintile 1, n=850, 13.3%).
Quality Indicators
Overall, the recording of quality indicators in patients with a history of stroke increased after the introduction of the new contract (Table 3), in particular, for the recording of cholesterol and smoking status. There was, however, only a small increase in the proportion of stroke patients achieving cholesterol and blood pressure control.
Sex Differences
Women had larger increases in the recording of quality indicators over time than did men. However, after the contract, women with a history of stroke were still less likely than men to have a recording of smoking status or having received antiplatelet or anticoagulant therapy, even after adjustment for age, number of stroke-related comorbidities, deprivation, and practice (Table 4). At both time points, women tended to be less likely to have controlled cholesterol and blood pressure than men.
Age Differences
There were large increases in quality indicator recording over time when the new contract was implemented, especially among the oldest patients (Table 5). The changes tended to reduce differences for particular quality indicators among age groups. After the contract, the oldest patients (>75 years) with a history of stroke were less likely than the youngest (<65 years) to have a recording of smoking status and cholesterol but more likely to have influenza vaccination recorded. The oldest patients were less likely to have received antiplatelet or anticoagulant therapy before the contract, a situation that was reversed after implementation of the new contract.
Deprivation Differences
In a comparison of care of patients in the highest deprivation category (quintile 5) with those in the least deprived group (quintile 1), patients in the least deprived category (quintile 1) had the largest increases in the recording of a magnetic resonance imaging/computed tomography scan (absolute difference, 47.0%), smoking (55.5%), cholesterol (53.1%), antiplatelet or anticoagulant therapy (36.9%), and influenza vaccination (36.8%) (Table 6). A significant difference between the most and least deprived patients emerged after the contract, with the most deprived stroke patients being less likely to have a record of smoking status and blood pressure.
Discussion
We found that the recording of data relating to the care of patients after stroke or TIA increased substantially during the period when Scottish practices were preparing for the new incentive-based contract. Similar improvements in aspects of care of chronic disease were also observed in 41 practices based in 6 areas of England in 2003, which saw improvements in the recording of symptoms and advice, in-house procedures, and test ordering.7
Large increases in the recording of risk factors in the oldest patients tended to attenuate age differences. Sex differences persisted in some components of care. More affluent patients tended to have larger increases in recording of quality indicators than did the most deprived. This resulted in increasing deprivation differences in certain aspects of care over time. One possible explanation for the oldest patients having the largest increases is the observation that older patients were more likely to have multiple comorbidities, some of which are included in other parts of the new contract (eg, diabetes and coronary heart disease). Patients with comorbidities may have been more likely to be reviewed by their GP in preparation for the new contract. This does not explain, however, why males and individuals in the most deprived group had the smallest increases in the recording of certain key components of the contract over time. A possible explanation for this result is that that patients in deprived areas and males may be less willing to seek advice for their condition. In addition, it has also been found that the average consultation length for deprived patients is &1 to 2 minutes shorter than for affluent patients.11 A reduced consultation time for deprived patients seeking advice from a GP may have reduced the opportunity for GPs to record risk factors.
Moderate increases in the attainment of cholesterol and blood pressure control were observed; however, the time interval of 1 year may not have been long enough to achieve significant improvements. In the long term, though, the new contract is likely to change the behavior of GPs as shown in previous schemes, which linked financial incentives with improved delivery of cervical smears and childhood immunizations.12 We will be able to monitor the practices to see whether outcomes improve over time.
After implementation of the new contract, there was a large increase in the number of patients with a recorded diagnosis of stroke or TIA. Although we did not have details of the criteria used to make the diagnosis of stroke, the increases in prevalence were probably the result of financial incentives for primary care practices to have accurate disease registers (with opportunities to exclude patients who refused or who were contraindicated treatment or who were too frail or refused to have clinical examinations). The rationale for use of a 15-month period was determined by the new GMS contract. It is possible that some overlap in data between the 2 time periods exists; however, the most recent recording (maximum date before the time point) of the quality measures was used to try and minimize this overlap. The likely consequence of this overlap is an underestimation of the increase in recording between the 2 time periods.
We were unable to control directly for disease severity, although we could allow for differences between groups at each time point by adjusting for a number of other confounders such as age, sex, deprivation, practice, and number of comorbidities. The data presented in this study are derived from the electronic recording of quality indicators on GP computer systems. The increase over time in quality indicator recording may have occurred partly because of the transfer of data from paper to electronic patient records (rather than from actual provision of additional care). It seems unlikely, however, that the practitioners would have systematically improved their computerized documentation of stroke-related care for only certain groups of patients (eg, the least deprived patients). The nonexperimental design of the study means that we cannot directly attribute the changes observed to the new contract. Other developments may have also contributed, although the strong relation between reaching targets and payment to practices suggests that the new contract was a major impetus for the changes.
Summary
Few studies have examined the effects of incentives offered to GPs for the provision of care.5 It appears that the introduction of the new contract has been associated with a dramatic rise in the recording of quality indicators, with moderate increases in the attainment of cholesterol and blood pressure control among patients with stroke. However, not all of the population seems to have benefited from these actions. This has important implications for these subgroups in terms of stroke recurrence and mortality. These data should inform strategies to improve the care for female, older, and more deprived stroke/TIA survivors in primary care to reduce the burden of stroke on the Scottish community.
Acknowledgments
The authors thank the Information Services Division of the Scottish Executive. The authors are grateful to the GPs who provided practice data to the Primary Care Clinical Informatics Unit.
Disclosures
None.
References
Weir N, Dennis MS. Towards a national system for monitoring the quality of hospital-based stroke services. Stroke. 2001; 32: 1415–1421.
Simpson CR, Wilson CHP, Hannaford PC, Williams D. Evidence for age and gender differences in the secondary prevention of stroke in Scottish primary care. Stroke. 2005; 36: 1771–1775.
The NHS Confederation. Investing in General Practice: The New General Medical Services Contract. http://www.bma.org.uk/ap.nsf/content/NewGMSContract. Last accessed October 2005.
Scottish Clinical Information Management in Primary Care group. Making It Work for You. http://www.scimp.scot.nhs.uk/gpg/doc_ page65.shtml. Last accessed October 2005.
Rosenthal M, Fernandopulle H, Song R, Landon B. Paying for quality: providers incentives for quality improvement. Health Aff. 2005; 23: 127–141.
Beaulieu N, Horrigan D. Putting smart money to work for quality improvement. Health Serv Res. 2005; 40: 1318–1334.
Campbell SM, Roland MO, Middleton E, Reeves D. Improvements in quality of clinical care in English general practice 1998–2003: longitudinal observational study. BMJ. 2005; 331: 1121–1123.
Helms PJ, Ekins-Daukes S, Taylor MW, Simpson CR, McLay JS. Utility of routinely acquired primary care data for paediatric disease epidemiology and pharmacoepidemiology. Br J Clin Pharmacol. 2005; 59: 331–341.
Whitelaw FG, Taylor RJ, Nevin SL, Taylor MW, Milne RM, Watt AH. Completeness and accuracy of morbidity and repeat prescribing records held on general practice computers in Scotland. Br J Gen Pract. 1996; 46: 181–186.
Carstairs V, Morris R. Deprivation and Health in Scotland. Aberdeen, Scotland: Aberdeen University Press; 1991.
Stirling AM, Wilson P, McConnachie A. Deprivation, psychological distress, and consultation length in general practice. Br J Gen Pract. 2001; 51: 456–460, 467.
Shekelle P. New contract for general practitioners. BMJ. 2003; 326: 457–458., 百拇医药(Colin R. Simpson, PhD; Philip C. Hannafo)
Abstract
Background and Purpose— We wished to ascertain whether a new contract based on financial incentives for general practitioners has been associated with improved recording of quality indicators for patients with stroke and whether there was evidence of any difference in change between sex, age, and deprivation groups.
Methods— In a serial cross-sectional study, patients from 310 general practices with a computer record of transient ischemic attack or stroke in Scotland were analyzed for their recording of quality indicators before and after the introduction of a new quality-based contract on March 31, 2004. Multivariate analyses were used to explore any differences in recording between age, sex, and deprivation groups.
Results— Documentation of quality indicators increased over time, with absolute increases for individual indicators ranging from 32.3% to 52.1%. There was a large increase in the documentation of quality indicators among the oldest patients (>75 years) and the most affluent patients. This tended to attenuate age groups differences and to exacerbate differences between deprivation groups. Women tended to have larger increases in documentation than men; however, sex differences persisted, with women less likely than men to have smoking habits recorded (adjusted odds ratio, 0.87; 95% confidence interval, 0.81 to 0.95) or to receive antiplatelet or anticoagulant therapy (adjusted odds ratio, 0.93; 95% confidence interval, 0.86 to 0.99).
Conclusions— The recording and management of quality indicators among patients with stroke increased substantially. However, inequitable care exists, which may have important implications for female, older, and more deprived subgroups in terms of stroke recurrence and mortality.
Key Words: epidemiology healthcare policy risk factors stroke stroke management
Introduction
Virtually all individuals resident in Scotland (including children) are registered with primary care, which is free at the point of contact and which provides first-line and continuing posthospitalization care of patients. Access to secondary care is usually obtained through a primary care practice, and even when a patient is admitted to hospital (eg, because of an emergency), details of the hospital stay are reported back to the patient’s primary care practice. So far, studies of stroke care have focused on hospital management, with very little work done in primary care.1 We were the first to show that primary care management of patients with stroke/transient ischemic attack (TIA) may be suboptimal.2
In April 2004, a quality-based general medical services (GMS) contract was introduced to Scottish general practice, which reduced the proportion of income of general practitioners (GPs) derived from per capita payments and increased the proportion (&23%) derived from providing specific aspects of care, such as targets based on quality indicators.3 The new contract provides payments to practices to develop an accurate register of patients who have had a stroke (because a complete and accurate register is a prerequisite for monitoring patients) and for the recording of smoking habits, blood pressure, and cholesterol levels of patients on the register. Further payments are made for reaching a number of specific treatment targets, eg, blood pressure control. The recording of quality indicators is based on the presence of disease rather than the demographic (eg, age, sex, or deprivation) characteristics of the population served. Initial payments were based on the achievement of targets 1 year after the introduction of the contract.
In April 2000, the Primary Care Clinical Informatics Unit was created as part of national primary care information management initiatives, including the Scottish Clinical Information Management in Primary Care and its Scottish Programme for Improving Clinical Effectiveness (SPICE).4 The role of the Primary Care Clinical Informatics Unit is to help GPs understand their clinical information needs through a variety of feedback reports based on their practices’ extracted data. SPICE care management screens are used by clinicians to systematically record data about a number of chronic diseases, including stroke. The data include information required for the new contract.3 Diagnostic criteria are not given; instead, diagnoses are recorded through routine practice, which, for major conditions like stroke, is often after investigation and input from hospital-based specialist colleagues.
Although few evaluations of incentive systems have been conducted,5 a recent study of chronic disease care carried out in New York suggested that financial incentives to primary care physicians leads to improvements in objective quality-of-care measures.6 Substantial improvements have also been seen in the quality of care provided by general practices for asthma, coronary heart disease, and type 2 diabetes as a result of systematic quality improvement initiatives in the United Kingdom.7
We hypothesized that during the period when the new quality-based contract was being prepared and introduced, the recording of computerized information about key components of care for patients who have had a stroke would improve. As a secondary aim, we also wanted to see whether there was any evidence of a change in differences between sex, age, and deprivation groups before and after institution of the new contract.
Subjects and Methods
Anonymous retrospective data from all 310 of the 850 Scottish practices that use the General Practice Administrative Software System and that participate in SPICE were obtained in November 2005. These 310 practices were self-selected; however, they have been shown to be representative of all Scottish practices.8 The completeness and accuracy of morbidity and repeat prescribing data in General Practice Administrative Software System practices have been reported previously.9 The long-term nature of the database and its clinical focus help ensure that initially uncertain events tend to be confirmed or refuted over time, especially those requiring continued care, like stroke. From the accumulated data, we identified everyone who had a computer record of a TIA (read codes G65 to G654, G656 to G65zz) or stroke (including cerebral hemorrhagic; read codes G61 below but not G617, G66, and below) and nonhemorrhagic stroke (read codes G63y0-1, G6760, G6w, G6x, G64, and below) on March 31, 2004 (1 year before introduction of the new contract in April 2004, designated as the "precontract" period in this article; total population at risk 1 806 266 patients) and March 31, 2005 (1 year after introduction of the new contract in April 2004; designated as the "postcontract" period; 1 775 397 patients). All registered patients with a recording of stroke before the 2 time points were included in the analyses.
Key characteristics of every identified person at each time point were determined, as follows: sex, age (<64, 65 to 75, or 75+ years), number of stroke-related comorbidities (diabetes [C10 and below], hypertension (G2, G20 and below, G24 to G2z], atrial fibrillation [G573 and below], coronary heart disease [G3 to G3401, G342 to G366., G38 to G3z], heart failure [G58 and below], and peripheral vascular disease [G73 and below]; (0, 1, 2, or 3+), and deprivation status based on postal code (Carstair’s DEPCAT categorization, which uses household overcrowding; unemployment, social class, and proportion of all persons in private households with no car as indicators of poverty,10 with deprivation quintile 1 as being least deprived to 5, being most deprived). We also determined for strokes occurring in the 15 months before each time point whether the event had been confirmed by computed tomography or magnetic resonance imaging by looking for documentation of such scans. In addition, for all stroke patients, we looked for a recording before the 2 time points for stroke-related quality indicators defined in the new contract: whether the person’s computer records contained details about the stroke diagnosis; smoking habits (current, ex-smoker, or never smoked) and (where appropriate) provision of smoking cessation advice; measurement and (where appropriate) control of cholesterol (to 5mmol/L) and blood pressure (to 150 mm Hg systolic and 90 mm Hg diastolic); provision of antiplatelet or anticoagulant therapy (including over-the-counter salicylate) to patients with nonhemorrhagic stroke or TIA; and provision of influenza vaccination.3 The time periods used for our data analyses were determined by the GMS contract. When 1 entry existed for a particular item, we used the most recent entry.
We excluded from the analysis anyone with a record of "exception codes," which indicate that the person has refused or been considered unsuitable to have any quality indicators measured. Patients were also excluded from particular analyses when an exception code pertaining to a quality indicator existed, eg, for aspirin treatment, if there was a recorded contraindication or allergy to this medicine.
Statistical Analysis
Binary logistic regression was used to determine odds ratios (ORs) and 95% confidence intervals (95% CIs) for the recording of quality indicators among different sex, age, and deprivation groups, adjusted for potential confounding by sex, age, number of stroke related comorbidities, deprivation, and practice (except where a factor was itself being explored). Patients with missing data (eg, smoking status) were excluded from the analysis of that factor. Where appropriate, 2 tests were used to compare differences between groups. For clarity, all proportions are presented to 1 decimal place and ORs, to 2 decimal places. All analyses were performed with SPSS for Windows 12.0 (SPSS Inc).
Results
By March 31, 2004 (precontract), 21 901 patients had a computer record of any stroke or TIA (1.2% of everyone registered with the practices; 95% CI, 1.1% to 1.2%). Forty-six patients had a computer record of an exception code. By March 31, 2005 (postcontract), the corresponding number of patients was 32 401 (1.8%; 95% CI, 1.8% to 1.8%), of whom 2565 had a computer record of an exception code. There was a greater proportion of female, older, and less deprived patients with stroke or TIA before the contract compared with after the contract. Stroke/TIA patients were also more likely to have a greater number of comorbidities recorded after the contract (Table 1).
After the contract, men were more likely to have diabetes, peripheral vascular disease, and coronary heart disease and less likely to have hypertension than women (Table 2). The oldest and most deprived patients were more likely to have a recorded diagnosis of all of the comorbidities analyzed. One exception was heart failure, for which there was no difference between the different deprived groups in this diagnosis.
Male, the oldest, and the most deprived patients tended to have 3 or more comorbidities recorded than women (men n=2621, 16.1%; women n=2159, 13.4%), the youngest (>75 years n=2647, 17.6%; <65 years n=634, 8.0%), and least deprived patients (deprivation quintile 5, n=749, 16.0%; deprivation quintile 1, n=850, 13.3%).
Quality Indicators
Overall, the recording of quality indicators in patients with a history of stroke increased after the introduction of the new contract (Table 3), in particular, for the recording of cholesterol and smoking status. There was, however, only a small increase in the proportion of stroke patients achieving cholesterol and blood pressure control.
Sex Differences
Women had larger increases in the recording of quality indicators over time than did men. However, after the contract, women with a history of stroke were still less likely than men to have a recording of smoking status or having received antiplatelet or anticoagulant therapy, even after adjustment for age, number of stroke-related comorbidities, deprivation, and practice (Table 4). At both time points, women tended to be less likely to have controlled cholesterol and blood pressure than men.
Age Differences
There were large increases in quality indicator recording over time when the new contract was implemented, especially among the oldest patients (Table 5). The changes tended to reduce differences for particular quality indicators among age groups. After the contract, the oldest patients (>75 years) with a history of stroke were less likely than the youngest (<65 years) to have a recording of smoking status and cholesterol but more likely to have influenza vaccination recorded. The oldest patients were less likely to have received antiplatelet or anticoagulant therapy before the contract, a situation that was reversed after implementation of the new contract.
Deprivation Differences
In a comparison of care of patients in the highest deprivation category (quintile 5) with those in the least deprived group (quintile 1), patients in the least deprived category (quintile 1) had the largest increases in the recording of a magnetic resonance imaging/computed tomography scan (absolute difference, 47.0%), smoking (55.5%), cholesterol (53.1%), antiplatelet or anticoagulant therapy (36.9%), and influenza vaccination (36.8%) (Table 6). A significant difference between the most and least deprived patients emerged after the contract, with the most deprived stroke patients being less likely to have a record of smoking status and blood pressure.
Discussion
We found that the recording of data relating to the care of patients after stroke or TIA increased substantially during the period when Scottish practices were preparing for the new incentive-based contract. Similar improvements in aspects of care of chronic disease were also observed in 41 practices based in 6 areas of England in 2003, which saw improvements in the recording of symptoms and advice, in-house procedures, and test ordering.7
Large increases in the recording of risk factors in the oldest patients tended to attenuate age differences. Sex differences persisted in some components of care. More affluent patients tended to have larger increases in recording of quality indicators than did the most deprived. This resulted in increasing deprivation differences in certain aspects of care over time. One possible explanation for the oldest patients having the largest increases is the observation that older patients were more likely to have multiple comorbidities, some of which are included in other parts of the new contract (eg, diabetes and coronary heart disease). Patients with comorbidities may have been more likely to be reviewed by their GP in preparation for the new contract. This does not explain, however, why males and individuals in the most deprived group had the smallest increases in the recording of certain key components of the contract over time. A possible explanation for this result is that that patients in deprived areas and males may be less willing to seek advice for their condition. In addition, it has also been found that the average consultation length for deprived patients is &1 to 2 minutes shorter than for affluent patients.11 A reduced consultation time for deprived patients seeking advice from a GP may have reduced the opportunity for GPs to record risk factors.
Moderate increases in the attainment of cholesterol and blood pressure control were observed; however, the time interval of 1 year may not have been long enough to achieve significant improvements. In the long term, though, the new contract is likely to change the behavior of GPs as shown in previous schemes, which linked financial incentives with improved delivery of cervical smears and childhood immunizations.12 We will be able to monitor the practices to see whether outcomes improve over time.
After implementation of the new contract, there was a large increase in the number of patients with a recorded diagnosis of stroke or TIA. Although we did not have details of the criteria used to make the diagnosis of stroke, the increases in prevalence were probably the result of financial incentives for primary care practices to have accurate disease registers (with opportunities to exclude patients who refused or who were contraindicated treatment or who were too frail or refused to have clinical examinations). The rationale for use of a 15-month period was determined by the new GMS contract. It is possible that some overlap in data between the 2 time periods exists; however, the most recent recording (maximum date before the time point) of the quality measures was used to try and minimize this overlap. The likely consequence of this overlap is an underestimation of the increase in recording between the 2 time periods.
We were unable to control directly for disease severity, although we could allow for differences between groups at each time point by adjusting for a number of other confounders such as age, sex, deprivation, practice, and number of comorbidities. The data presented in this study are derived from the electronic recording of quality indicators on GP computer systems. The increase over time in quality indicator recording may have occurred partly because of the transfer of data from paper to electronic patient records (rather than from actual provision of additional care). It seems unlikely, however, that the practitioners would have systematically improved their computerized documentation of stroke-related care for only certain groups of patients (eg, the least deprived patients). The nonexperimental design of the study means that we cannot directly attribute the changes observed to the new contract. Other developments may have also contributed, although the strong relation between reaching targets and payment to practices suggests that the new contract was a major impetus for the changes.
Summary
Few studies have examined the effects of incentives offered to GPs for the provision of care.5 It appears that the introduction of the new contract has been associated with a dramatic rise in the recording of quality indicators, with moderate increases in the attainment of cholesterol and blood pressure control among patients with stroke. However, not all of the population seems to have benefited from these actions. This has important implications for these subgroups in terms of stroke recurrence and mortality. These data should inform strategies to improve the care for female, older, and more deprived stroke/TIA survivors in primary care to reduce the burden of stroke on the Scottish community.
Acknowledgments
The authors thank the Information Services Division of the Scottish Executive. The authors are grateful to the GPs who provided practice data to the Primary Care Clinical Informatics Unit.
Disclosures
None.
References
Weir N, Dennis MS. Towards a national system for monitoring the quality of hospital-based stroke services. Stroke. 2001; 32: 1415–1421.
Simpson CR, Wilson CHP, Hannaford PC, Williams D. Evidence for age and gender differences in the secondary prevention of stroke in Scottish primary care. Stroke. 2005; 36: 1771–1775.
The NHS Confederation. Investing in General Practice: The New General Medical Services Contract. http://www.bma.org.uk/ap.nsf/content/NewGMSContract. Last accessed October 2005.
Scottish Clinical Information Management in Primary Care group. Making It Work for You. http://www.scimp.scot.nhs.uk/gpg/doc_ page65.shtml. Last accessed October 2005.
Rosenthal M, Fernandopulle H, Song R, Landon B. Paying for quality: providers incentives for quality improvement. Health Aff. 2005; 23: 127–141.
Beaulieu N, Horrigan D. Putting smart money to work for quality improvement. Health Serv Res. 2005; 40: 1318–1334.
Campbell SM, Roland MO, Middleton E, Reeves D. Improvements in quality of clinical care in English general practice 1998–2003: longitudinal observational study. BMJ. 2005; 331: 1121–1123.
Helms PJ, Ekins-Daukes S, Taylor MW, Simpson CR, McLay JS. Utility of routinely acquired primary care data for paediatric disease epidemiology and pharmacoepidemiology. Br J Clin Pharmacol. 2005; 59: 331–341.
Whitelaw FG, Taylor RJ, Nevin SL, Taylor MW, Milne RM, Watt AH. Completeness and accuracy of morbidity and repeat prescribing records held on general practice computers in Scotland. Br J Gen Pract. 1996; 46: 181–186.
Carstairs V, Morris R. Deprivation and Health in Scotland. Aberdeen, Scotland: Aberdeen University Press; 1991.
Stirling AM, Wilson P, McConnachie A. Deprivation, psychological distress, and consultation length in general practice. Br J Gen Pract. 2001; 51: 456–460, 467.
Shekelle P. New contract for general practitioners. BMJ. 2003; 326: 457–458., 百拇医药(Colin R. Simpson, PhD; Philip C. Hannafo)