Perceptions of self-rated health among stroke survivors: a qualitative study in the United Kingdom.

BACKGROUND: Self-rated health predicts health outcomes independently of levels of disability or mood. Little is known about what influences the subjective health experience of stroke survivors. Our aim was to investigate stroke survivors' perceptions of self-rated health, with the intention of informing the design of interventions that may improve their subjective health experience. METHODS: We conducted semi-structured interviews with a purposive sample of 28 stroke survivors recruited from a stroke unit and follow-up outpatient clinic, 4-6 months after stroke, to explore what factors are perceived to be part of self-rated health in the early stages of recovery. Qualitative data were analysed using a thematic analysis approach to identify underlying themes. RESULTS: Participants' accounts show that stroke survivors' perceptions of self-rated health are multifactorial, comprising physical, psychological and social components. Views on future recovery after stroke play a role in present health experience and are shaped by psychosocial resources that are influenced by past experiences of ill-health, dispositional outlook such as degree of optimism, a sense of control and views on ageing. CONCLUSIONS: Severity of physical limitations alone does not influence perceptions of self-rated health among stroke survivors. Self-rated health in stroke survivors is a multidimensional construct shaped by changes in health status occurring after the stroke, individual characteristics and social context. Understanding the factors stroke survivors themselves associate with better health will inform the development of effective approaches to improve rehabilitation and recovery after stroke.

Introducing structured caregiver training in stroke care: findings from the TRACS process evaluation study.

OBJECTIVE: To evaluate the process of implementation of the modified London Stroke Carers Training Course (LSCTC) in the Training Caregivers After Stroke (TRACS) cluster randomised trial and contribute to the interpretation of the TRACS trial results. The LSCTC was a structured competency-based training programme designed to help develop the knowledge and skills (eg, patient handling or transfer skills) essential for the day-to-day management of disabled survivors of stroke. The LSCTC comprised 14 components, 6 were mandatory (and delivered to all) and 8 non-mandatory, to be delivered based on individual assessment of caregiver need. DESIGN: Process evaluation using non-participant observation, documentary analysis and semistructured interviews. PARTICIPANTS: Patients with stroke (n=38), caregivers (n=38), stroke unit staff (n=53). SETTINGS: 10 of the 36 stroke units participating in the TRACS trial in four English regions (Yorkshire, North West, South East and South West, Peninsula). RESULTS: Preparatory cascade training on delivery of the LSCTC did not reach all staff and did not lead to multidisciplinary team (MDT) wide understanding of, engagement with or commitment to the LSCTC. Although senior therapists in most intervention units observed developed ownership of the LSCTC, MDT working led to separation rather than integration of delivery of LSCTC elements. Organisational features of stroke units and professionals' patient-focused practices limited the involvement of caregivers. Caregivers were often invited to observe therapy or care being provided by professionals but had few opportunities to make sense of, or to develop knowledge and stroke-specific skills provided by the LSCTC. Where provided, caregiver training came very late in the inpatient stay. Assessment and development of caregiver competence was not commonly observed. CONCLUSIONS: Contextual factors including service improvement pressures and staff perceptions of the necessity for and work required in caregiver training impacted negatively on implementation of the caregiver training intervention. Structured caregiver training programmes such as the LSCTC are unlikely to be practical in settings with short inpatient stays. Stroke units where early supported discharge is in place potentially offer a more effective vehicle for introducing competency based caregiver training. LINKED TRACS CLUSTER RANDOMISED CONTROLLED TRIAL NUMBER: ISRCTN49208824.

An observational study of acute stroke care in four countries: the European registers of stroke study.

BACKGROUND: The European Registers Of Stroke (EROS) project aimed to assess outcomes of stroke care across Europe, relating these to both case mix information from disease-specific population registers and the quality of stroke care provided at each centre. This included comparing information on quality of care with direct observation of the stroke care process in 4 centres. METHODS: Direct non-participant observational methods were used on a purposive sample of first-stroke patients admitted within the past 14 days to an acute-stroke unit or ward that admits stroke patients in 4 urban hospital sites in London, UK, Dijon, France, Kaunas, Lithuania, and St. Petersburg, Russia. We recorded patient characteristics with levels of contact with multi-disciplinary team (MDT) members and contact with families and mobilization to build a collection of 'snapshots' of stroke care throughout the patients' day. One independent observer undertook all observations over 1 day. RESULTS: We observed differences between centres in the proportion of observations where patients were alone (lowest proportion in London, highest proportion in St. Petersburg) (p > 0.001), where patients had contact with MDT members (p > 0.001) and family, and where patients were out of bed/mobile (p > 0.001) (both with highest proportion in London, lowest proportion in St. Petersburg). CONCLUSIONS: Higher levels of contact with the MDT, family contact and mobilization were observed in the Western European centres than the Eastern European and Russian centres. Differences in case mix may explain some, but not all, of these differences. Direct observation has some limitations; however, it could be developed in future studies to help identify other key aspects of effective stroke care.

Sociodemographic differences in return to work after stroke: the South London Stroke Register (SLSR).

BACKGROUND: Loss of employment contributes significantly to the burden of stroke on individuals and society. There is limited information on factors influencing return to work after stroke. OBJECTIVES: To investigate the frequency and determinants of return to paid work after stroke in a multi-ethnic urban population. METHODS: Patterns of return to work were examined among people with first ever stroke registered in the population based South London Stroke Register. Employment status and functional outcome (Barthel Index (BI), Frenchay Activity Index (FAI)) were assessed 1 year after stroke. Associations between baseline characteristics and return to paid work were analysed by multivariable logistic regression analysis. RESULTS: Among 2874 patients with first ever strokes in 1995-2004, 400 (15%) were working before the stroke. At 1 year, 94 (35%) of 266 survivors had returned to paid work. Black ethnicity (OR 0.41; 95% CI 0.19 to 0.88), female sex (0.43; 0.21 to 0.91), older age (p<0.001), diabetes (0.25; 0.08 to 0.79) and dependence (BI < or = 19) in the acute phase (0.24; 0.11 to 0.49) were independently associated with lower odds of return to work in multivariable analysis. Better functional outcome at 1 year was associated with return to paid work (p<0.001) but 53% of 161 independent (BI > 19) and 39% of 96 very active (FAI > 30/45) individuals had not resumed work. CONCLUSIONS: There were important sociodemographic differences in return to work after stroke that were independent of clinical and service use variables included in the analysis. A large proportion of patients did not resume work despite excellent functional outcome.

Involving older people in health research.

BACKGROUND: It is a UK policy requirement to involve patients and the public in health research as active partners. OBJECTIVE: We reviewed published reports of studies which involved older people in commissioning, prioritizng, designing, conducting or disseminating research. Search strategy and selection criteria: systematic searches of databases (PubMed, SCI-EXPANDED, SSCI, A&HCI, ASSIA, Embase, CINAHL and Medline) for English language studies published between 1995 and 2005 which had involved older people as partners n the research process as opposed to research subjects. Articles were reviewed by two authors using a standardised matrix for data extraction. RESULTS: Thirty studies were included and classified according to the stage in the research process in which older people were involved. Barriers to involving older people were: cultural divisions, language barriers, research skills capacity, ill health, time and resources. Four of the studies had been formally evaluated to identify the impact of involvement. Evaluation focussed on the impact on participants rather than on impact on research processes and outcomes. Benefits to participants included: increased knowledge, awareness and confidence, meeting others in similar situations, empowering older people to become active in their community regarding decisions/policies which affect them. CONCLUSIONS: Factors hindering the involvement of older people in research were the same as reported factors hindering involvement of younger people, suggesting that age, per se, is not a barrier. To demonstrate the impact of user involvement on research quality, the definition of user involvement requires clarification, and systematic evaluation of research involving older people needs to be developed.

Clinical determinants of long-term quality of life after stroke.

OBJECTIVES: To determine factors that independently predict health-related quality of life (HRQOL) 1 and 3 years after stroke. METHODS: Subjects numbering 397, from a population-based register of first-ever strokes were assessed for HRQOL using the Short Form 36 (SF36) 1 year after stroke. Physical (PHSS) and mental health (MHSS) summary scores were derived from the eight domains of HRQOL in the SF36. Multivariate stepwise regression analyses were conducted to determine independent predictors of these scores; beta coefficients with 95% CI were obtained.beta coefficient is the difference between average value of the variable (e.g. male) and average value under consideration (e.g. female). Demographic and stroke risk factors, neurological impairments and cognitive impairment (MMSE <24) were included in the models. Similar analyses were undertaken on 150 subjects 3 years post-stroke. RESULTS: A year after stroke, independent predictors of the worst PHSS were of females (beta coefficient -3.3: 95% CI -5.7 to -0.8), manual workers (-3.2: -5.9 to -0.4), diabetes (-4.2: -7.7 to -0.8), right hemispheric lesions (-4.9: -8.7 to -1.2), urinary incontinence (-7.8: -11.6 to -4.1) and cognitive impairment (-2.7: -5.5 to -0.1); the worst MHSS were associated with being Asian (-11.8: -20.6 to -3.0), ischaemic heart disease (-2.7: -5.4 to -0.03), cognitive impairment (-3.04: -5.8 to -0.3). Subjects aged 65-75 years (5.4: 2.5 to -8.4) had better MHSS than those <65 years. Three years post-stroke, independent predictors of worse PHSS were hypertension (-8.7: -13.5 to -3.9), urinary incontinence (-8.1: -15 to -1.1) and cognitive impairment (-8.3: -13.2 to -3.5). CONCLUSIONS: Determinants of HRQOL vary both over time after stroke and whether physical or psychosocial aspects of HRQOL are being considered. This study provides valuable information on factors predicting long-term HRQOL, which can be taken into consideration in audits of clinical practice or in future interventional studies aiming to improve HRQOL after stroke.

Relationships between long-term stroke disability, handicap and health-related quality of life.

OBJECTIVES: To estimate levels of disability, handicap and health-related quality of life (HRQOL) up to 3 years after stroke and examine the relationships between these domains. DESIGN: A longitudinal, observational study SETTING: Population-based register of first-ever strokes METHODS: Subjects, registered between 1 January 1995 and 31 December 1997, were assessed at 1 year (n = 490) and 3 years (n = 342) post-stroke for disability [Barthel index (BI)], handicap [Frenchay activity index (FAI)] and HRQOL (SF-36). BI was categorised as severe, moderate, mild and independent (0-9, 10-14, 15-19 and 20); FAI was categorised as inactive, moderately active and very active (0-15, 16-30 and 31-45). SF-36 domains include: Physical Functioning (PF), Role Physical (RP), Bodily Pain (BP), General Health (GH), Vitality (VT), Social Functioning (SF), Role Emotional (RE) and Mental Health (MH). Physical (PHSS) and Mental Health (MHSS) Summary Scores were computed. RESULTS: at 1 and 3 years, 26.1 and 26.3%, respectively, were disabled (BI < 15); 55 and 51%, respectively, were handicapped (FAI = 0-15); and survivors had low mean PHSS (37.1 and 37.9), but satisfactory mean MHSS (46.6 and 47.7). There was a graded positive relationship between all SF-36 domains and the categories of BI and FAI. Spearman rank correlations were significant between BI and all SF-36 domains at both time points: strong (r > 0.70) with PF, moderate (r = 0.31-0.70) with RP, SF and PHSS, but weak (r < 0.30) with other domains. Correlations between FAI and SF-36 domains were strong with PF, weak with BP, RE and MHSS, and moderate with other domains. CONCLUSIONS: Disability and handicap remain highly prevalent up to 3 years after stroke. Patients' perception of physical health is persistently low, but mental health perception is satisfactory up to 3 years. Due to variable correlations between different HRQOL domains with disability and handicap, it is suggested that disability, handicap and HRQOL should all be assessed to acquire a broader measure of stroke outcome.

Investigating ethnic differences in sexual health: focus groups with young people.

OBJECTIVES: To compare normative beliefs about sexual health in young men and women from black Caribbean, black African, and white ethnic groups in order to better understand ethnic inequalities in sexual health. METHODS: Focus group discussions with young people living in an area with known high prevalence of gonorrhoea and chlamydia. Groups were stratified by sex and self defined ethnicity. RESULTS: 22 male and 20 female 16-25 year olds of black Caribbean, black African, and white ethnicity took part in eight discussions. Participants from black ethnic groups were more aware of gonorrhoea than white participants but all ethnic groups regarded these as being less important than unplanned pregnancy or HIV/AIDS. Most participants believed that they would have obvious symptoms if they had a sexually transmitted infection and could determine the cleanliness of sexual partners by visual or behavioural cues. Black Caribbean women were alone in acknowledging the likelihood of their partners having concurrent sexual relationships. Some black Caribbean women described negative attitudes of staff in genitourinary medicine clinics who were from the same ethnic background. CONCLUSION: In this focus group study we identified ethnic differences in terminology, awareness of sexually transmitted infections, non-exclusive sexual relationships, and experience of sexual health services but gender had a greater influence on normative beliefs. The similarities in norms for all ethnic groups might reflect common social and cultural exposures. The low priority given to sexually transmitted infections by young people from all ethnic groups needs to be addressed if they are to be tackled successfully.

Defining and using quality of life: a survey of health care professionals.

OBJECTIVE: 'Quality of life' is an important but poorly defined outcome in health and health care research. We sought to identify stroke professionals' definitions of quality of life and views of the purpose of its assessment. DESIGN: Using issues identified during in-depth interviews with stroke care professionals, we designed a postal survey questionnaire. Participants were asked to define quality of life, identify the purposes of assessing it and report experiences of measuring patient quality of life. Comparisons between professional groups were analysed using chi-squared tests of significance. SUBJECTS: Care of the elderly physicians, physiotherapists and occupational therapists in the UK. RESULTS: Of 2793 questionnaires distributed, 1572 were returned (56% response rate). Quality of life was defined in terms of 'happiness' by 72%; 25% included social well-being; 25% included physical abilities. Most (91%) identified 'asking patients' as an effective way to assess quality of life; 40% using standardized measures. Half those who reported using quality of life measures experienced difficulties, including being unsure about which measure to use and concerns about validity. CONCLUSIONS: The idea of quality of life as happiness dominated responders' definitions. We argue that the term may be used in both a technical sense (an outcome) and in a broad colloquial sense, without necessarily distinguishing between the two. Clarification of the concept and its uses is required if recent calls to introduce quality of life assessment in clinical care are to be feasible.

'You can't tell by looking': pilot study of a community-based intervention to detect asymptomatic sexually transmitted infections.

Innovative and non-stigmatizing interventions are required to reduce ethnic inequalities in rates of sexually transmitted infections among young people. We therefore designed an intervention, 'You can't tell by looking,' which combined health promotion with testing for gonorrhoea and chlamydia using nucleic acid amplification technology and treatment and partner notification delivered in the non-clinical settings. One hundred and eighty-one participants were seen in 13 sessions in local further education colleges. Forty-three percent of participants were from Black Caribbean or Black Other ethnic groups and 39% were Black African: 125 of 181 participants were sexually active and 109 of these (87%) provided a urine specimen. 10/109 (9.2%, 95% confidence interval 4.5-16.2%) samples were confirmed positive for Chlamydia trachomatis and two were also positive for Neisseria gonorrhoeae. Only 7% of those tested found it embarrassing. The intervention was both feasible and acceptable to young people. It could be tested in a wider variety of non-clinical settings and evaluated in a cluster randomized trial.

Two simple questions to assess outcome after stroke: a European study.

BACKGROUND AND PURPOSE: The "2 simple questions" were designed as an efficient way of measuring outcome after stroke. We assessed the sensitivity and specificity of this tool, adapted for use in 8 European centers, and used it to compare outcomes across centers. METHODS: Data were taken from the Biomed II prospective study of stroke care and outcomes. Three-month poststroke data from 8 European centers were analyzed. Sensitivity and specificity were assessed by comparing responses to the 2 simple questions with Barthel Index and modified Rankin scale scores. Adjusting for case mix, logistic regression was used to compare patients in each center with "good" outcome (not dependent and fully recovered) at 3 months. RESULTS: Data for 793 patients were analyzed. For the total sample, the dependency question had a sensitivity of 88% and a specificity of 77%; the recovery question had a sensitivity of 78% and a specificity of 90%. Dependency data from Riga had much lower sensitivity. There was variation in good outcome between centers (P:=0.0015). Compared with the reference center (Kaunas), patients in Dijon, Florence, and Menorca were more likely to have good outcome, after adjusting for case mix. CONCLUSIONS: Dependency and recovery questions showed generally high sensitivity and specificity. There were significant differences across centers in outcome, but reasons for these are unclear. Such differences raise particular questions about how patients interpreted and answered the simple questions and the extent to which expectations of recovery and perceived needs for assistance vary cross-culturally.

Behavioral risk factor prevalence and lifestyle change after stroke: a prospective study.

BACKGROUND AND PURPOSE: Stroke patients have a 15-fold increased risk of recurrent stroke, and those with > or =1 risk factor have a further increased risk of recurrence. Previous work found management of physiological risk factors after stroke to be unsatisfactory, but there is little information on behavioral risks within the stroke population. This study estimates behavioral risk factor prevalence after stroke and explores lifestyle change. METHODS: The study used data from the population-based South London Stroke Register, collected prospectively between 1995 and 1998. Main measures included smoking status, alcohol use, and obesity. Logistic regression was used to determine sociodemographic differences in these measures. RESULTS: At 1 year after stroke, 22% of patients still smoked, 36% of patients were obese, and 4% drank excessively. Younger patients, whites, and men were more likely to smoke, and younger whites were more likely to drink excessively. Women and nonwhites were more likely to be obese. Those living in hospital, nursing home, or residential care and nonwhites were more likely to give up smoking, but there were no other associations between lifestyle change and the sociodemographic characteristics of patients. CONCLUSIONS: Different behavioral risk factors were associated with specific sociodemographic groups within the stroke population. After stroke, high-risk groups should continue to be targeted to prevent stroke recurrence. However, the relationship between sociodemographic characteristics and lifestyle change remains unclear; more research is needed into the process of change to find out how best to intervene to improve secondary prevention.

GPs' employment of locum doctors and satisfaction with their service.

BACKGROUND: Locum doctors provide cover during normal working hours for GPs absent due to holidays, sickness, maternity leave or for educational purposes. However, there is little information on the extent of the use of locums or of GPs' perception of their services. OBJECTIVES: To examine the level of use of locum doctors by GPs, the ease of recruitment and satisfaction with their services. METHODS: A postal survey of all general practices in one of the six health regions in England was carried out. Logistic regression analysis was used to examine the independent effects on locum use of practice size and type of area, source of recruitment and GPs' satisfaction with their services. RESULTS: A total of 935 (80.6%) general practices responded. Locum GPs were employed by 81.7% of practices in the previous 12 months. Two-thirds of practices reported problems obtaining locum cover, especially at short notice and for holiday periods. One-fifth of practices employing a locum in the previous 12 months were dissatisfied with the locum. CONCLUSIONS: There are high demands for, but a considerable shortage of, locum doctors in general practice. Educational and other initiatives for GPs may contribute to increased demands for locum cover. Difficulties in recruitment may be reduced by measures to improve the conditions of employment for doctors working as locums on a longer term basis. New codes of practice for employing locums may increase satisfaction with locum services.

Locum doctors in general practice: motivation and experiences.

BACKGROUND: There is evidence of dissatisfaction with locum doctors' performance, but little is known about doctors who work as locums in general practice or about their experiences of this work. AIM: To describe the motivations and experiences of doctors providing locum cover in general practices. METHOD: A postal questionnaire survey distributed to locums through organizations such as locum groups, commercial agencies, and general practices. RESULTS: Questionnaires were returned by 111 doctors currently working as locums in general practice. Four main reasons for working as a locum GP were: as a short-term option while between posts, to gain experience of different practices before commitment to one practice, to balance work and family or other commitments, to continue part-time work after retirement. One-quarter of responders intended to continue working as a locum indefinitely. The drawbacks of locum work included frustration with low status, lack of security, and difficulty accessing structured training and education. CONCLUSION: Locum doctors in general practice are a heterogeneous group that includes those who have chosen this type of work. The doctors who intend to continue as locums indefinitely represent a useful resource in primary care whose ability to provide short-term cover could be maximized. The need to control the quality of 'freelance' doctors should not overshadow the need to control the quality of their working environments.

Sickness absence and 'working through' illness: a comparison of two professional groups.

BACKGROUND: Few studies have investigated occupational groups reporting low rates of sickness absence because of an assumption that these rates indicate low morbidity. This is inconsistent with the view that sickness absence, which may be caused by social and psychological rather than medical factors, does not equate with morbidity. This paper investigates rates of sickness absence and factors influencing decisions not to take sick leave among doctors and a comparative professional group. METHODS: A postal survey was sent to 670 general practitioners (GPs), 669 hospital doctors and 400 company 'fee earners'. Qualitative interviews were conducted with 64 doctors reporting an illness lasting one month or more in the last three years. RESULTS: Self-reported health status was similar for both groups but GPs reported higher levels of occupational stress. However, doctors were significantly less likely to report short periods of sick leave in the previous year. Over 80 per cent of all respondents had 'worked through' illness, citing cultural and organizational factors behind their decision not to take sick leave. Barriers to sick leave among doctors included the difficulty of arranging cover and attitudes to their own health. CONCLUSIONS: Considerable emphasis has been given to the role of social factors in contributing to rates of sickness absence. These may also contribute to the decision not to take sick leave, resulting in possible inappropriate non-use. Measures to encourage and enable doctors to take sick leave might improve the management of their own health.

Illness doesn't belong to us.

There has been little research in Britain into the experiences of doctors who are ill. We conducted in-depth interviews with 64 doctors of all grades with a recent illness lasting one month or more. Whether the illness was physical or psychiatric, many expressed the idea that illness is inappropriate for doctors. This idea is a cultural value among doctors which is reinforced by the organization of medical work. It discourages doctors from seeking and obtaining appropriate help when they are ill.

Changing use of hospital services and costs at a London AIDS referral centre, 1983-1989.

OBJECTIVE: To describe the use of hospital-based services and associated costs over time for HIV-infected individuals by disease stage and year of treatment. METHODS: Data on service use were abstracted from inpatient and outpatient case-notes of 459 HIV-infected patients (121 asymptomatic and 338 AIDS patients) treated at St Mary's Hospital, London between 1983 and 1989. Cost estimates were derived from a survey of the 37 departments involved with HIV-related care. The number and type of admissions and outpatient visits, referral and discharge venues, number and type of tests and procedures performed, drugs prescribed and associated costs for the study period were the outcome measures used. RESULTS: Most patients were homosexual men. At the time of first HIV-related visit, 80.7% lived in London. Most day cases and planned admissions were organized through the outpatient clinic; 31% of emergency admissions were initiated by patients themselves. For people with AIDS the number of day case admissions increased while planned admissions decreased. There was a marked reduction in the duration of inpatient stays, especially for AIDS patients. Costs associated with inpatient care decreased concomitantly. The number of outpatient visits for patients with symptomatic disease increased, resulting in increased outpatient expenditure. Asymptomatic patients had fewer inpatient tests, while outpatient tests did not change over time; costs followed similar patterns. Mean inpatient and outpatient drug-days prescribed did not change nor did average inpatient drug-costs although outpatient drug-costs increased. Inpatient tests performed on symptomatic non-AIDS patients decreased, while mean outpatient tests increased; average costs followed similar patterns. Inpatient drug-days prescribed and costs remained the same, while outpatient drug-days and average drug-costs increased during the study period. For AIDS patients, the number of inpatient tests performed and their average costs decreased but outpatient tests performed increased, though their average costs remained the same. Mean inpatient drug-days prescribed and average drug-costs decreased, while number of outpatient drug-days prescribed and average drug-costs increased markedly over time. For each disease category, expenditure on admissions and related tests decreased over time, while expenditure on outpatient visits and drug-costs increased. CONCLUSIONS: The shift from an inpatient- to an outpatient-based service has resulted in fewer patients being investigated and treated in hospital and more as outpatients. This has resulted in a reduction of inpatient-related costs, while outpatient-related costs have increased. The overall contribution of drug-costs to the total cost has increased greatly over time. With the anticipated advent of new antiviral compounds, the importance of ascertaining their effectiveness as well as their efficacy will become crucial.

Changing presentation and survival, service utilization and costs for AIDS patients: insights from a London referral centre.

OBJECTIVE: To describe the use of services and associated costs for HIV-infected patients by stage of infection in the context of changing patterns of presentation and survival. METHODS: A retrospective analysis of inpatient and outpatient case-notes of 335 AIDS patients and a survey of HIV-related care provided by 37 departments at St Mary's Hospital, London. Survival from time of diagnosis of AIDS, time from diagnosis of HIV infection to AIDS, and use and costs of services per patient-year were outcome measures. RESULTS: During the study period 1 January 1982 to 30 September 1989, 152 AIDS patients were diagnosed before 1987 (group 1) and 183 since 1987 (group 2), most of whom were homosexual men. The median interval of first HIV-related visit to diagnosis of AIDS increased from 0 (group 1) to 264 days (group 2; P < 0.0001). Median survival from AIDS diagnosis increased from 14.6 (group 1) to 21.0 months (group 2; P < 0.02). Group 2 patients used fewer inpatient services than group 1 patients irrespective of disease stage. Symptomatic patients in group 2 used more outpatient services than group 1 patients. Total HIV-related expenditure was lower for patients without AIDS in group 2 than in group 1, while expenditure for AIDS patients remained similar. CONCLUSION: Earlier patient presentation, a shift from inpatient- to outpatient-based clinical care and increased survival from time of AIDS diagnosis has occurred. Increased drug expenditure was offset by reduced inpatient expenditure. Total expenditure per patient-year was stable; increased survival and introduction of new drugs will increase future lifetime use of resources.