An Exploratory Analysis of Referrals to a major Emergency Department in a developing country

Background: Hospital crowding, ED waiting times and high demand for unscheduled care all place significant burdens on secondary care services. This impacts on patient care, staff morale and overall functioning of the whole healthcare system. Patient referrals from other healthcare providers often is a result of limited access to resources, specialists or because of acuity. However, some referrals may be more suitable for lower acuity settings, with the benefit of better overall patient experience. In addition, duplication of contacts with a healthcare professional may not result in additional benefit to patients, but may necessarily add to the patient journey and contribute to crowding. Objectives: We aimed to determine the originator of referrals to the ED. We also aimed to determine the proportion of referred patients who received any meaningful intervention at the ED. Finally, we aimed to estimate the proportion of patients referred who may have been suitable for direct inpatient referral or management in a lower acuity setting. Methods: We conducted a prospective evaluation of all referrals to the ED of a large urban hospital over 7 days. Routine anonymised demographic, diagnosis and intervention data were collected and simple descriptive analysis was undertaken using Microsoft Excel®. A validated algorithm was applied to determine suitability for lower acuity settings, and contextual secondary analysis was applied to determine choice of altResults: There were 168 formal referrals during the period evaluated (mean 24/day), of which data was available for 151. Most referrals were on Monday and Thursday. 39.7% were referred from the four regional District Health Facilities (DHF). 12 % were referred by specialists. There were significantly higher referrals from Local Health Centres located more than 5km of the hospital compared with those closer, although this could have been due to greater numbers outside the 5 km radius. 5.5% were thought suitable for primary care management and 31% could have been referred directly to an inpatient team if this were available. The majority (51.3%) of referred patients received no significant intervention in the ED, with almost 1 in 7 suitable for outpatient management. Conclusions: A significant number of patients referred to the ED may have been more appropriately directed. Direct special admission, access to outpatient referral slots or telephone advice from senior ED or specialty clinicians may prevent up to a half of referrals being seen by an ED clinician. This may reduce unnecessary transport, improve time and resource utilization and decongest the ED and hospital. Further large scale evaluation is warranted to investigate the predictors of referral, control for seniority, and make more robust recommendations for improving the patient journey ernate pathways.

Changing patterns of primary care for diabetes in Trinidad and Tobago over 10 years.

OBJECTIVE: To evaluate standards of preventive medical care for Type 2 diabetes in the context of high prevalence and limited resources. METHODS: Surveys of records for diabetic subjects attending 23 government primary care health centres in Trinidad and Tobago in 2003 and 1998 and nine health centres in 1993. Records were compared by study year for blood glucose and blood pressure, surveillance for complications, appropriate management advice and drug prescriptions. Relevant interventions included new clinical guidelines, training workshops for clinical staff and reports to the Ministry of Health. During this time, investment in primary care increased in the context of health sector reform policies and a favourable macroeconomic environment. RESULTS: Comparing 1993 with 2003, the proportion with a blood glucose test in the past 12 months increased from 33% to 91%, urea or creatinine ever recorded increased from 14% to 61%, diet advice recorded in the first 12 months of follow-up from 35% to 67%, exercise advice from 3% to 61%. The proportion prescribed metformin increased from 25% to 65%, while chlorpropamide decreased from 48% to 0%. The proportion of all subjects treated with antihypertensive drugs increased from 49% to 70%, and the proportion of treated patients prescribed angiotensin-converting enzyme inhibitors increased from 8% to 72%. Most recent ever records of blood glucose, blood pressure and body weight showed no decrease. CONCLUSIONS: Repeated surveillance of processes of care provided information to stimulate and plan change. Process changes were associated with intervention at several levels and increased availability of resources.

Diabetes-related inequalities in health status and financial barriers to health care access in a population-based study.

AIMS: We evaluated the relationship between diabetes, health status, household income and expenditure on health care in the general population in Trinidad. METHODS: Multistage sampling of 300 households was used to select a sample of 548 adults aged > or = 25 years. There were 64 (12%) who reported a diagnosis of diabetes. Comparison was made with 128 non-diabetic controls who were frequency matched for age and sex. RESULTS: Subjects with diabetes had lower income levels than non-diabetic controls [income < or = US dollars 533 per month for 66% diabetes cases and 48% controls, test for trend P = 0.007]. Compared with controls, subjects with diabetes were less likely to have good or very good self-rated health (diabetes 32%, controls 67%; P < 0.001), and more frequently reported long-standing illness, limitation of activities, visual impairment, or self-reported history of high blood pressure, angina or heart attack. Subjects with diabetes (11%) were less likely than controls (30%) to have private health insurance (P = 0.005). Diabetic subjects (35%) were more likely than controls (16%) to have incurred expenditure on doctors' services in the last 4 weeks (P = 0.021). CONCLUSIONS: Diabetes is associated with worse health status and more frequent expenditure on medical services but greater financial barriers to access in terms of low income and lack of health insurance. Policies for diabetes should specifically address the problem of income-related variations in risk of diabetes, health care needs and barriers to uptake of preventive and treatment services, otherwise inequalities in health from this condition may increase.

Socioeconomic inequality in blood pressure and its determinants: cross-sectional data from Trinidad and Tobago.

We evaluated income- and education-related inequalities in blood pressure, hypertension and hypertension treatment in the general population of Trinidad and Tobago. The design included survey of 300 households in north central Trinidad, including 631 adults in 2001. Measurements of blood pressure, weight, height, waist and hip circumferences, and educational attainment, household income and alcohol intake by questionnaire. The slope index of inequality (SII) was used to estimate the difference in blood pressure between those with highest, as compared to lowest, socioeconomic status. Complete measurements and questionnaires were obtained for 461 (73%) including 202 men and 259 women. In women, after adjusting for age and ethnicity, the SII for systolic blood pressure by income was -12.6, 95% confidence interval -22.6 to -2.6 mmHg (P=0.013); and -10.8 (-21.4 to -0.2) mmHg (P=0.045) by educational attainment. After additionally adjusting for body mass index, waist-hip circumference ratio and self-reported diabetes, the SII for income was -7.3 (-16.5 to 1.9) mmHg (P=0.120) and for educational attainment was -3.0 (-13.0 to 6.9) mmHg (P=0.551). In men, after adjusting for age and ethnicity, the SII for systolic blood pressure by income was -4.3 (-15.4 to 6.8) mmHg (P=0.447) and for education -8.1 (-19.0 to 2.8) (P=0.145). There is a negative association of systolic blood pressure with increasing income or education in women. This is associated with body mass index, abdominal obesity and diabetes. There is no consistent association between education or income and blood pressure in men.

Social inequalities and children's height in Trinidad and Tobago.

OBJECTIVE: The aim of the study was to report the association of socio-economic factors with child's height. DESIGN: Cross-sectional study based on a representative national sample of government schools. SETTING: Trinidad and Tobago in 1999. SUBJECTS: A total of 2608 boys and girls mean age 5.8 y, range 4.38-6.99 y and 3080 mean age 8.6 y, range 7.00-10.44 y olds. OUTCOME: Measurement of height and a questionnaire completed by parents. In the analysis height was expressed as standard deviation scores (s.d.s.) based on the British height curves (1990) or height below -1.5 s.d.s. RESULTS: Ethnicity, parental heights, birthweight, maternal age at child's birth and number of children in the family were the main factors associated with children's height. Lack of piped water supply in the home was the only socio-economic factor consistently associated with height (mean difference in s.d.s. adjusted only for age group, gender and ethnicity -0.192, 95% CI -0.257 to -0.127 and in addition adjusted for the variables listed above -0.080, 95% CI -0.141 to -0.019). Parental education, household overcrowding and employment status were weakly associated with height in the partially adjusted model only. Analysis of severe growth failure gave similar results. CONCLUSION: The impact of socio-economic factors on height is marginal in Trinidad and Tobago. As socio-economic factors may have an impact on a broad range of health indicators, height and rates of undernutrition should not be used as sole criteria for assessing progress in decreasing health differentials caused by social inequalities.

Free school meals and children's social and nutritional status in Trinidad and Tobago.

OBJECTIVE: To evaluate the provision of free school meals in Trinidad and Tobago in relation to children's social and nutritional status. DESIGN AND METHODS: Cross-sectional survey of a nationally representative sample of 66 government schools, including children in the admissions classes (aged 4 to 7 years) and classes for 'rising nines' (aged 7-10 years). Data included questionnaire details of free school meals and children's social background, and measurements of children's heights, weights and skinfold thicknesses. RESULTS: Of 6731 eligible children, data were analysed for 5688 (85%). There were 2386 (42%) children receiving free meals provided at school. At different schools the proportion of all children receiving free meals ranged from 20% to 100%, Receipt of free meals was associated with larger family size (one child, 32% received free meals; > or =6 children, 63%), lower paternal educational attainment (primary, 52% free; university, 30%), father's employment (employed, 39% free meals; unemployed >12 months, 59%) as well as maternal education and employment and household amenities. After adjusting for age, sex and ethnic group, children who received free meals were shorter (mean difference in height standard deviation score (SDS) -0.12, 95% confidence interval (CI) -0.17 to -0.06), lighter (body mass index SDS -0.21, -0.28 to -0.14) and thinner (subscapular skinfold SDS -0.13, -0.18 to -0.09). CONCLUSIONS: Free school meals were widely available, with some targeting of provision to children with less favourable social and nutritional status. Greater universality would reduce inequity, but more stringent targeting and reduction of school-level variation would increase efficiency.

Overweight, obesity and skinfold thicknesses of children of African or Indian descent in Trinidad and Tobago.

BACKGROUND: To evaluate distribution of body mass index (BMI) and subcutaneous fat in children of African or Indian subcontinent descent living in Trinidad and Tobago. METHODS: A cross-sectional survey was carried out in a nationally representative sample of 66 government schools. Measurements were made of children's heights, weights, triceps and subscapular skinfold thicknesses. Data were analysed by sex and ethnic group and comparison was made with international standards for overweight and obesity and with British (1990) reference curves for BMI. RESULTS: Data were analysed for 5688/6731 (85%) eligible children including 1934 Afro-Trinidadian, 1689 Indo-Trinidadian, and 1794 of mixed ethnicity. Afro-Trinidadian and mixed children were taller, but Indo-Trinidadian children were shorter than reference. Values for BMI were lower than reference: mean standard deviation score (SDS), (95% confidence interval) Afro-Trinidadian -0.34 (-0.39 to -0.28), Indo-Trinidadian -1.32 (-1.39 to -1.25), mixed -0.67 (-0.74 to -0.61). Overall 486 (8.5%, 7.8 to 9.3%) of children were overweight and 138 (2.4%, 2.0 to 2.9%) were obese. Triceps skinfold thickness values were lower than reference (-0.45, -0.48 to -0.42 SDS) but subscapular skinfold thicknesses were higher (0.45, 0.43 to 0.47 SDS). Higher BMI were associated with higher BMI in the child's parents, higher reported birthweight, older age of the child's mother, smaller family size, and with higher maternal educational attainment. CONCLUSIONS: Overweight was prevalent and distribution of subcutaneous fat was central. Work is needed to determine whether these findings are associated with adult patterns of fat distribution and metabolic abnormalities.

Overweight, obesity and skinfold thicknesses of children of African or Indian descent in Trinidad and Tobago

BACKGROUND: To evaluate distribution of body mass index (BMI) and subcutaneous fat in children of African or Indian subcontinent descent living in Trinidad and Tobago. METHODS: A cross-sectional survey was carried out in a nationally representative sample of 66 government schools. Measurements were made of children's heights, weights, triceps and subscapular skinfold thicknesses. Data were analysed by sex and ethnic group and comparison was made with international standards for overweight and obesity and with British (1990) reference curves for BMI. RESULTS: Data were analysed for 5688/6731 (85%) eligible children including 1934 Afro-Trinidadian, 1689 Indo-Trinidadian, and 1794 of mixed ethnicity. Afro-Trinidadian and mixed children were taller, but Indo-Trinidadian children were shorter than reference. Values for BMI were lower than reference: mean standard deviation score (SDS), (95% confidence interval) Afro-Trinidadian -0.34 (-0.39 to -0.28), Indo-Trinidadian -1.32 (-1.39 to -1.25), mixed -0.67 (-0.74 to -0.61). Overall 486 (8.5%, 7.8 to 9.3%) of children were overweight and 138 (2.4%, 2.0 to 2.9%) were obese. Triceps skinfold thickness values were lower than reference (-0.45, -0.48 to -0.42 SDS) but subscapular skinfold thicknesses were higher (0.45, 0.43 to 0.47 SDS). Higher BMI were associated with higher BMI in the child's parents, higher reported birthweight, older age of the child's mother, smaller family size, and with higher maternal educational attainment. CONCLUSIONS: Overweight was prevalent and distribution of subcutaneous fat was central. Work is needed to determine whether these findings are associated with adult patterns of fat distribution and metabolic abnormalities.

Utilisation of private care by public primary care clinic attenders with diabetes: relationship to health status and social factors.

Private health care provision is important in most middle-income countries but factors influencing the demand for private care have not been well defined. This paper evaluated the relationships of health status and socio-economic variables with utilisation of private care by public primary care clinic attenders. The sample included 2117 randomly selected subjects with clinical diabetes attending 35 government health centres in Trinidad and Tobago. Measures included attendance at a private doctor, the type 2 Diabetes Symptom Checklist, the SF36 questionnaire, and indicators of socio-economic status. Of the sample, 1256 (59%) reported attending a private doctor, 577 (27%) attended a private doctor for diabetes, and 378 (18%) attended a private doctor regularly. Attendance at a private doctor was associated with lower SF36 scores. The odds ratio for a 10 unit increase in SF36 physical component score was 0.81, 95% confidence interval 0.72-0.91. After adjusting for demographic and social factors the relative odds were 0.89, 0.80 to 1.00. After allowing for differences in health status, the relative odds of attending for private care for those without a pipe borne water supply in the home, compared with those with water in the home, were 0.77, 0.63-0.94. Those living alone were less likely to attend a private doctor than those living with their children and partner (odds ratio 0.60, 0.43-0.83). Among people attending public clinics, the decision to utilise private care is sensitive to health status. After adjusting for health status, there was evidence for horizontal inequity in access to private care in relation to household amenities and composition.

Evaluating variations in medical practice between government primary care health centres.

We evaluated variations in medical care for diabetes among government health centres in Trinidad and Tobago. Data were analysed for 1579 subjects with clinical diabetes attending 23 health centres concerning 12 processes of care and six case-mix or outcome variables. Random effects models were used for analysis. Health centre level variation was appreciable. Intraclass correlations ranged from 0.025 to 0.316 for process of care variables and 0.000 to 0.056 for case-mix variables. Compared with health centres with only one nurse, patients attending those with three or more nurses received 1.18 (95% confidence interval 0.27 to 2.09) more processes of care. Delivery of medical care varied among the health centres and those with higher staffing levels provided more processes of care.

A five-year evaluation of intervention in diabetes care in Trinidad and Tobago.

AIMS: To evaluate an intervention to improve diabetes care in government-run heath centres in Trinidad and Tobago over 5 years. METHODS: A cross-sectional survey of 690 subjects with clinical diabetes attending nine health centres was carried out in 1993. The intervention was: reports to the Ministry of Health, dissemination of management guidelines and annual training workshops for healthcare staff. Re-evaluation was through a survey of 1579 subjects with diabetes, attending 23 health centres in 1998. RESULTS: Comparing 1993 with 1998, foot examinations in the previous year increased from 38 (6%) to 346 (22%) and fundoscopy from 6 (1%) to 139 (9%). For subjects attending for 1 year or less, 34/96 (35%) had dietary advice recorded in 1993 compared with 77/143 (54%) in 1998. Exercise advice was recorded for 3/96 (3%) in 1993 and 48/143 (34%) in 1998. In 1993, 329 (48%) were taking chlorpropamide but this fell to 57 (4%) in 1998. Glibenclamide use increased from 214 (31%) to 856 (54%) and gliclazide from four (1%) to 205 (13%). In 1993, 198/338 (56%) of hypertensive subjects were taking Brinerdin, this fell to 56/829 (7%) in 1998 while use of thiazide diuretics, methyldopa and angiotensin-converting enzyme (ACE) inhibitors increased. There were no changes in indicators of metabolic control, blood pressure control or body weight. CONCLUSIONS: Use of audit data to inform health policy and practice, linked with educational interventions, may modify patterns of care in government-run primary care health centres in a middle-income country with a high prevalence of diabetes.

A 4-year evaluation of blood pressure management in Trinidad and Tobago.

This paper reports a 4-year evaluation of government primary care services in Trinidad and Tobago. The sample included 16 primary care health centres in Trinidad and Tobago with cross-sectional surveys of adult attenders in 1994 and 1998. Data were obtained from clinic records including details of processes of care, drug utilisation and blood pressure (BP) control. Intervention following the initial survey was by means of reports to the Ministry of Health and annual training workshops for medical officers. Data were analysed for 1759 adult attenders at study clinics (981 in 1994 and 778 in 1998). Most attenders had blood pressure recorded (96% in 1994 and 98% in 1998). Among 1176 subjects with hypertension (BP > or =160/95 mm Hg or treated with drugs) the proportion with dietary advice ever recorded increased from .139/662 (21%) in 1994 to 185/514 (36%) in 1998, and recording of exercise advice increased from 36 (5%) in 1994 to 99 (19%) in 1998. Among subjects treated with drugs, use of Brinerdin decreased from 191/625 (31%) in 1994 to 29/486 (6%) in 1998, while use of diuretics, beta-blockers, calcium antagonists and ACE inhibitors increased. The proportion of patients with blood pressure adequately controlled (<160/95 mm Hg) was 338 (51%) in 1994 and 297 (58%) in 1998 (odds ratio 1.39, 95% confidence interval 0.96 to 2.00). An audit study used to inform health care policy, staff training and clinical practice may contribute to improving processes of hypertension management in a middle-income country like Trinidad and Tobago.

Relationship of health-related quality of life to symptom severity in diabetes mellitus: a study in Trinidad and Tobago.

Our objective was to estimate the effect of greater symptom severity in diabetes mellitus on measures of health-related quality of life in a cross-sectional design in 35 government primary care health centres in Trinidad. Data were gathered on 2,117 subjects with clinical diabetes and analysed for 1,880 (89%). For each scale of the short form 36 (SF-36) questionnaire (a generic measure of health-related quality of life), scores were presented by quartile of symptom severity, measured using the Diabetes Symptom Checklist. Mean (SD) SF-36 scores were 44 (10) for the physical component score (PCS) and 45 (12) for the mental component score (MCS). Greater severity of diabetic symptoms was associated with lower scores on each of the subscales of the SF-36. Comparing lowest and highest quartiles of DSC score, the adjusted difference in PCS was -11 (95% confidence interval -12 to -9) and for MCS -16 (-18 to -14). Our results provide standardised data for health related quality of life in relation to severity of illness from diabetes, these might be used to aid the evaluation of relevant interventions.

Acute severe asthma in Trinidad and Tobago.

SETTING: Accident and emergency department, General Hospital, Port of Spain, Trinidad and Tobago. OBJECTIVE: To measure the occurrence and clinical management of acute severe asthma. DESIGN: Data collected prospectively for consecutive attendees with asthma over a period of 3 months. RESULTS: Asthma accounted for 8.8% of attendances. Treatment for the acute attack included nebulised salbutamol in 1031 (85%) and nebulised atrovent in 769 (63%). Systemic corticosteroids were given to 623 (51%) patients. Only 247 (20%) had planned follow-up arrangements recorded. Usual maintenance treatment included inhaled salbutamol in 767 (63%) and inhaled corticosteroid in 286 (24%). Inhaled corticosteroids were more often used by patients aged > or =15 years or who had had previous hospital admissions for asthma. CONCLUSION: The survey identified deficiencies in the clinical management of acute asthma attacks and in longterm asthma care. Caribbean guidelines for asthma care have subsequently been published, and follow-up surveys should be carried out to evaluate their implementation.

Evaluation of a home kit blood glucose monitoring device

A blood glucose monitoring device, the Diascan, is commonly used in Trinidad and Tobago. A prospective study was conducted to examine the accuracy of a Diascan unit in measuring blood glucose levels in or capillary venous blood of patients in a hospital ward. The Diascan measurements were compared to those from two laboratories which independently measured the venous blood or the venous plasma glucose levels. Although there was reasonably good correlation between measurements from the two laboratories (r = 0.85) results from the Diascan showed poor correlations with those from the laboratories, with Pearson's correlation coefficients ranging from 0.32 to 0.64. An error grid analysis showed that the Diascan measurements would have resulted in inappropriate decisions relating to treatment regimens in 26of cases. The results suggest that, when crucial decisions have to be made with respect to patients' blood glucose levels, it may be risky to rely solely on measurements from the Diascan.

Social inequalities in morbidity from diabetes mellitus in public primary care clinics in Trinidad and Tobago.

Associations between socio-economic status and non-communicable diseases in middle income countries have received little study. We conducted an interview survey to evaluate the associations of morbidity with social conditions among people attending government primary care health centres with diabetes mellitus in Trinidad. Data collected included morbidity from hyperglycaemia, foot problems, visual problems and cardiovascular disease, as well as social and demographic variables. Of 622 subjects, 35% were aged > or = 65 years, 54% were Indo-Trinidadian, 13% had no schooling, only 11% were in full-time employment, and 33% had no piped drinking water supply in the home. Prevalent symptoms included itching, reported by 215 (35%), nocturia in 315 (51%), burning or numbness in the feet in 350 (56%), and difficulty with eyesight in 363 (58%). A morbidity summary score was used as dependent variable in regression analyses. Comparing those with no schooling with those with secondary education, the mean difference in morbidity score was 1.77 (95% CI 1.15-2.39), attenuated to 0.71 (0.06-1.37) after adjusting for age, gender, ethnic group and diabetes duration. The equivalent differences for those with no piped water supply in the house, compared with those with, were 0.53 (0.17-0.88) and 0.57 (0.24-0.89). For the unemployed, compared with those in full-time jobs, at ages 15-59 years the differences were 0.85 (0.14-1.56) and 0.58 (-0.11-1.27). We conclude that morbidity in persons with diabetes is associated with indicators of lower socio-economic status and that this association is partly explained by confounding with older age, female gender, longer duration of diabetes and Indo-Trinidadian ethnic group. A negative association between socio-economic status and morbidity from diabetes contributes to a justification for investment of public health resources in the control of diabetes and other non-communicable diseases.

Evaluation of a home kit blood glucose monitoring device.

A blood glucose monitoring device, the Diascan, is commonly used in Trinidad and Tobago. A prospective study was conducted to examine the accuracy of a Diascan unit in measuring blood glucose levels in or capillary venous blood of patients in a hospital ward. The Diascan measurements were compared to those from two laboratories which independently measured the venous blood or the venous plasma glucose levels. Although there was reasonably good correlation between measurements from the two laboratories (r = 0.85) results from the Diascan showed poor correlations with those from the laboratories, with Pearson's correlation coefficients ranging from 0.32 to 0.64. An error grid analysis showed that the Diascan measurements would have resulted in inappropriate decisions relating to treatment regimens in 26% of cases. The results suggest that, when crucial decisions have to be made with respect to patients' blood glucose levels, it may be risky to rely solely on measurements from the Diascan.

Stroke in Trinidad and Tobago: burden of illness and risk factors.

This study describes the burden of stroke on hospital services in a Caribbean community. The settings are the two main acute general hospitals in Trinidad observed over a 12-month period. All subjects were admitted with a clinical diagnosis of acute stroke. The measures were hospital admission rates, length of hospital stay, case-fatality rates, disability at discharge, and risk factors for stroke. There were 1,105 hospital admissions with a diagnosis of stroke. The median length of stay was 4 days, with an interquartile range of 2 to 9, and stroke accounted for approximately 9,478 bed days per annum. The hospital admission fatality rate was 29%. Among surviving patients, 437 (56%) were severely disabled at discharge. Age-standardized admission rates for first strokes in persons aged 35-64 years were 114 (95% CI: 83 to 145) per 100,000 in Afro-Trinidadian men and 144 (109 to 179) in Indo-Trinidadian men. The equivalent rates for women were 115 (84 to 146) and 152 (118 to 186). Among patients with first strokes, 348/531 (66%) reported physician-diagnosed hypertension, but only 226 (65%) of these reported being on antihypertensives at admission. Stroke in Trinidad and Tobago is associated with a high case-fatality rate and severe disability in survivors. Modifiable risk factors were reported in a majority of stroke cases, and there is a need to develop effective preventive strategies.

Medical practitioners' views on the management of hypertension in Trinidad and Tobago.

We surveyed 161 medical practitioners in Trinidad and Tobago (124 reporting private sector practice and 37 describing government health centre practice) for their views on blood pressure (BP) management. 96% of the respondents agreed that BP should be measured on all adults seen and 90% agreed that diastolic pressure should be recorded as the disappearance of sounds. There was disagreement over the level of diastolic BP at which drug treatment should be initiated: 63% would treat diastolic BP less than 100 mm Hg, but 35% would only treat diastolic BP of 110 mm Hg or higher. In private practice 31% preferred angiotensin converting enzyme (ACE) inhibitors as treatment for an African Caribbean man with diastolic BP 110 mm Hg, but in public clinics 41% preferred thiazide diuretics. ACE inhibitors were most often preferred as treatment for an Indo-Caribbean man with diabetes and diastolic BP 110 mm Hg in both public and private practice. Doctors considered that noncompliance (66%), lack of education (34%) and unhealthy lifestyles (25%) were important obstacles to BP control. In private practice doctors considered patients' financial constraints to be an obstacle (58%), whereas in the public sector limited availability of drugs (57%) was felt to be more important. Less costly and, possibly, more appropriate drugs were used in public clinics.