A systematic review of evidence on the association between hospitalisation for chronic disease related ambulatory care sensitive conditions and primary health care resourcing
© Gibson et al.; licensee BioMed Central Ltd. 2013
Received: 11 April 2013
Accepted: 20 August 2013
Published: 26 August 2013
Primary health care is recognised as an integral part of a country’s health care system. Measuring hospitalisations, that could potentially be avoided with high quality and accessible primary care, is one indicator of how well primary care services are performing. This review was interested in the association between chronic disease related hospitalisations and primary health care resourcing.
Studies were included if peer reviewed, written in English, published between 2002 and 2012, modelled hospitalisation as a function of PHC resourcing and identified hospitalisations for type 2 diabetes as a study outcome measure. Access and use of PHC services were used as a proxy for PHC resourcing. Studies in populations with a predominant user pay system were excluded to eliminate patient financial barriers to PHC access and utilisation. Articles were systematically excluded based on the inclusion criteria, to arrive at the final set of studies for review.
The search strategy identified 1778 potential articles using EconLit, Medline and Google Scholar databases. Ten articles met the inclusion criteria and were subject to review. PHC resources were quantified by workforce (either medical or nursing) numbers, number of primary care episodes, service availability (e.g. operating hours), primary care practice size (e.g. single or group practitioner practice—a larger practice has more care disciplines onsite), or financial incentive to improve quality of diabetes care. The association between medical workforce numbers and ACSC hospitalisations was mixed. Four of six studies found that less patients per doctor was significantly associated with a decrease in ambulatory care sensitive hospitalisations, one study found the opposite and one study did not find a significant association between the two. When results were categorised by PHC access (e.g. GPs/capita, range of services) and use (e.g. n out-patient visits), better access to quality PHC resulted in fewer ACSC hospitalisations. This finding remained when only studies that adjusted for health status were categorised. Financial incentives to improve the quality of diabetes care were associated with less ACSC hospitalisations, reported in one study.
Seven of 12 measures of the relationship between PHC resourcing and ACSC hospitalisations had a significant inverse association. As a collective body of evidence the studies provide inconclusive support that more PHC resourcing is associated with reduced hospitalisation for ACSC. Characteristics of improved or increased PHC access showed inverse significant associations with fewer ACSC hospitalisations after adjustment for health status. The varied measures of hospitalisation, PHC resourcing, and health status may contribute to inconsistent findings among studies and make it difficult to interpret findings.
KeywordsType 2 diabetes Chronic disease Primary health care resourcing Ambulatory care sensitive conditions Hospitalisation
Primary health care (PHC) has been established as an integral part of a country’s health care system . Key functions of PHC as described in the Ottawa Charter include: keeping people and populations healthy; free from contractible illness and disease; providing timely treatment of treatable acute episodes of illness; and managing existing health conditions, in particular chronic conditions.
It is well documented that the natural progression of many chronic conditions results in multiple and complex morbidity . When chronic disease is managed well disease progression may be slowed and further morbidity prevented . Managing chronic diseases in the PHC setting is promoted through the use of evidence-based care guidelines that require a multi-disciplinary team approach to care . For this reason, the Australian Government provides monetary incentives to general practitioners (GPs) to prepare a care plan for each of their patients with a chronic illness .
To evaluate how well a PHC service is performing, hospitalisations are identified that could be avoided with accessible and high quality PHC. Such hospitalisations are commonly termed Ambulatory Care Sensitive Conditions (ACSC) or avoidable hospitalisations. The initial set of ACSC were defined in the United States in 1976 and has since been used as indicators of access to and the performance of PHC in Australia and elsewhere [7, 8].
Subsequently studies investigating access to PHC services, using ACSC as the outcome measure, were conducted in population groups facing different out-of-pocket costs in accessing PHC [9, 10]. Weissman et al.  compared hospitalisation rates for ACSC in health care insured and uninsured populations and found uninsured patients had higher rates of ACSC than those with health insurance. The implication of what Weissman et al.  report in terms of ACSC in insured and uninsured patients is that removing financial barriers to access is important for better outcomes. It is recognised that even by removing the direct financial cost of care other barriers to accessing care may exist. Other barriers to accessing chronic disease management include lack of time to return to appointments, lack of health education and unavailable suitable transport . In countries that have universal health care coverage, with subsidized or free access to PHC, such as in Australia, the United Kingdom and Spain, ACSC hospitalisation rates have been used primarily as an indicator of PHC quality (incorporating access) [12–14].
A set of ACSC has not been agreed upon universally. Generally though, the following categories of hospital conditions are considered to be ambulatory care sensitive; influenza and pneumonia, other vaccine preventable diseases, asthma, congestive heart failure, diabetes complications, chronic obstructive pulmonary disease, angina, iron deficiency anaemia, hypertension, nutritional deficiencies, dehydration and gastroenteritis, pyelonephritis, perforated/bleeding ulcer, cellulitis, pelvic inflammatory disease, ear nose and throat infections, dental conditions, convulsions and epilepsy, gangrene .
Symptom management (health status)
Supportive relationships (supportive spouse, house overcrowding)
Self-management support (personal resilience)
Coordination of care between primary, secondary and tertiary services (hospital discharge planning)
Local area liveability (air pollutants, geographical location)
Socio-economic opportunities (employment, health insurance)
How were hospitalisation and PHC resource variables measured?
Was the association between PHC resourcing and diabetes related hospitalisations significant, and if so, did the level of PHC resourcing increase or decrease hospitalisations?
What patient health risks or community factors, like socio-economic status, were associated with diabetes-related hospitalisations?
A search strategy was used to identify published studies that potentially addressed our research questions. Three database platforms were searched. EbscoHost platform was used to search the EconLit database. Search terms used were ‘hospital* OR ‘avoidable hospital*’ OR ‘ambulatory care sensitive’ AND ‘primary care’ OR ‘primary health care’ AND diabetes OR ‘type 2 diabetes’ OR ‘diabetes mellitus’. OvidSP platform was used to search the Medline database. Search terms were hospitalisation and primary health care and the Medline related words (for example hospitalisation, hospitalized) were applied. These were searched for in all fields (e.g. title, abstract) and across all geographic regions. Google scholar search engine was used with the term, ‘primary care resource and avoidable diabetes related hospitalisations’. The search was limited to English language, peer reviewed articles, and articles published between 2002 and 2012. References cited in articles considered eligible at the abstract stage were included in the review process.
Studies that used diabetes-related ACSC hospitalisations as at least one outcome measure of PHC performance or hospitalisation of persons with T2DM.
PHC resource variable(s) were quantified and included in the final model.
Studies in populations with a public health insurance scheme that financially covered all or the majority of health care costs. This was to exclude studies in populations where financial access barriers (high out-of-pocket costs for health care) would be a dominant influence on use of PHC and hospitalisations; given our research question was concerned with the impact of PHC resourcing in the context of universal health care access.
Studies published between 2002 and 2012, to ensure currency of health care systems and contemporary evidence for chronic care in PHC settings.
A systematic exclusion process was employed to arrive at the final set of review articles. From the initial search result article titles were reviewed. If it was obvious from the title that the study did not meet the inclusion criteria and was not peer reviewed (e.g. government report), the article was excluded.
no adjustment for confounders; individual or population health risk factors or community level characteristic that may affect hospitalisation, for example level of education
articles that could not be retrieved.
Studies excluded at the full article review stage a
Exclusion criteria at article review
Reference(notes for references)
Not a diabetes-related hospitalisation or a type 2 diabetes cohort
Did not measure a primary health care resource input
Did not adjust for patient level health risk of hospitalisation or social and economic factors that influence hospitalisation
Not a peer reviewed journal article
Analysis combining hospital outcomes and primary care inputs into a regional efficiency measure
Total articles excluded on full article review
Data was extracted from the final set of articles and presented in tables.
Diabetes-related hospital outcome measures and primary health care resource inputs and direction of significant study findings
First author, date published (country)
Hospital outcome measure
Primary health care resource inputs and direction of significant study findings
Category of diabetes-related ACSC (reporting of results)
Description of how measured (level of variable)
PHC resources significantly associated with an increase [↑] or decrease [↓] in hospitalisationa
(level of variable)
PHC variables that were not significant i.e. p value >0.05, or reference measure (level of variable)
Dusheiko 2011 (England)
Emergency (unplanned) hospitalisations due to (all) short-term diabetic complicationsb (incidence rate)
Incidence rate per family practice (health centre) (f)
Population per FTE family physician (f)
Griffiths 2010 (England)
Non-elective diabetes-related hospitalisations (rate per facility)
Rate per number of patients on the register experiencing ≥ 1 hospitalisation (f)
Increase in the number of patients per FTE GP(f) [↓]c <3038 patients per FTE practice nurse (f) [↑] 3039–3901 patients per FTE practice nurse (f) [↑] 4823–6210 patients per FTE practice nurse (f) [↓] 6210+ patients per FTE practice nurse (f) [↓]
Sole practitioner practice (f) Primary medical service contract (f) 3901–4823 patients per FTE practice nurse i.e. Quintile 3 (f)
Rate per number of patients on the register experiencing ≥ 2 hospitalisation (f)
Number of patients per FTE GP (f) [↓] <3038 patients per FTE practice nurse (f) [↑] 3901–4823 patients per FTE practice nurse (f) [↑]
Sole practitioner practice (f) Primary medical service contract (f) 3039–3901 patients per FTE practice nurse i.e. Quintile 2 (f) 4823–6210 patients per FTE practice nurse (f) i.e. Quintile 4 6210+ patients per FTE practice nurse (f) i.e. Quintile 5
Lavoie 2010 (Canada)
Chronic ACSC hospitalisation (rate difference)
Average difference in rates of hospitalisation between level of primary care serviced (f)
Health centre versus no facility (f) [↓] Health office versus no facility (f) [↓] Health centre versus nursing station (f) [↑] Health office versus nursing station (f) [↑]
Nursing station and no facility (f)
Ng 2010 (Canada)
An acute hospitalisation for any reason among persons age 12 years or older with type 2 diabetes (odds ratio)
Status of hospitalisation (yes or no) (i)
An increase in self-reported number of GP contacts in the previous 12 months [↓]
Bruni 2009 (Italy)
Hyperglycemic emergency hospitalisations (probability of being hospitalised)
Hospitalised, yes or no (i)
As number of visits to diabetes outreach clinic increased (i) [↑] More patients per gp 1100–1500 and >1500 (iGP) [↑] Larger proportion of annual income from pay-for-participation (GP payments related to number of patients with diabetes) (iGP) [↓] Health district receives ≥75 % GP income from incentive schemes (d) [↓]
Patients per GP <1100 (ref) Practice type, i.e. sole practitioner (ref), association, network, group (iGP) Per cent diabetic patients (iGP) Per cent annual income pay-for-compliance (GP payments related to the number of quality improvement processes involved in e.g. diabetes audit) (iGP) Health district receives 25–75 % GP income from incentives schemes (d)
El-Din 2009 (Saudi Arabia)
Type 2 diabetes related hospitalisation (odds of being hospitalised)
Hospitalised, yes or no (i)
≥ 6 outpatient PHC clinic visits, except diabetes clinic (i) [↑]
No outpatient clinic visits (ref) (i) 1–5 outpatient clinic visits (i)
Lin 2009 (Taiwan)
Short-term diabetes ACSC and long-term ACSC modelled separately (relative risk ratio)
Status of hospitalisation (yes or no) (i)
More outpatient diabetes visits per year (i) [↑]
Diabetes management received (primary care clinic (ref), medical centre, regional or district hospitals (i)
Rizza 2007 (Italy)
Hospitalisation for diabetes ambulatory care sensitive conditions (odds ratio)
Status of hospitalisation (yes or no) (i)
As the number of patients per primary care physician increases (iGP) [↑]
Number of primary care physician visits in previous year (i) Number of specialist visits in community health services (f)
Gulliford 2004e (England)
Hospitalisation for chronic conditions (chronic hospital admissions per 100 000 persons)
Rate of hospitalisation per 100 000 persons (ha)f
As GP supply increases per 10 000 weighted population (ha) [↓] As mean partnership size increases (ha) [↓] As proportion of sole provider practices increase (ha) [↑]
Per cent practices with diabetes service (ha)
Gulliford 2002e (England)
Hospitalisation for chronic conditions (chronic hospital admissions per 100 000 persons)
Rate of hospitalisation per 100 000 persons (ha)
As GP supply increases per 10 000 persons (ha) [↓]
Description of the final model including diabetes-related hospitalisation predictor variables, other than primary health care resourcing
First author and date published (country)
Type of analysis (reporting of results)
Study design (date)
Health risks and socio-economic factorsasignificantly associated with an increase [↑] or decrease [↓] in hospitalisationb(level of variable)
Independent variables that were not significantcor reference (level of variable)
Variance explained by the model (r-squared)
Dusheiko 2011 (England)
Negative binomial regression (incident rate ratio)
Prospective open cohort (2001/02 to 2006/07)
% HbA1c ≤7.4/7.5 (f) [↓] % 7.4/7.5 < HbA1c ≤10 (f) [↓] % HbA1c monitored (f) [↑] Baseline hospitalisation rate (f) [↑] Average physician age (f) [↓] % non-principal [↓]physicians (f) Training practice (f) [↑] % females 15–44 & 75+ years (f) [↑] Diabetes prevalence (f) [↑] Mental health prevalence (f) [↑] Heart disease prevalence (f) [↓] COPD prevalence (f) [↑] Low income index (f) [↑] % smoking (c) [↑] % obese (c) [↑] % communal residents (c) [↓] Located urban sparse, village/hamlet, village/hamlet sparse [↓] Mean distance to nearest practice (c)[↑]
Practice population (f) Personal medical services contract practice(f) % female physicians (f) % UK qualified physicians (f) % males all age groups (f) % females by age group other than age 15–44 & 75+ years(f) % non-white (c) % incapacity benefit (c) % binge drinking (c)Education/qualification deprivation (c) Central heating deprivation (c) Crime (c) Urban location (ref) Located town and fringe and fringe sparse (f) Mean distance to nearest 5 hospitals (f)
Efron’s R2 = 0.206
Griffiths 2010 (England) [outcome is ≥ 1 or ≥ 2 diabetes admissions]
Two-level multilevel model with GP practices nested within Primary Care Trusts (hospitalisation rate from count of admissions)
Cross sectional (2005/06)
Index of deprivation (f) [↑] % aged ≥65 years (f) [↓] % ethnic minority (f) [↓]
Least deprived (ref) Density (people per hectare) (f) GP ≥45 years (f) % female GPs (f) % GP qualified in UK (f)
[outcome is standardised diabetes admission ratio]
Density (people per hectare) (f) [↑] Unadjusted T2DM prevalence (f) [↑] % female GP (f) [↓] % GPs UK qualified (f) [↓]
% ethnic minority (f) GP ≥45 years (f)
Lavoie 2010 (Canada)
Generalised estimating equations (average difference in ACSC hospitalisation rates among different facility types)
Prospective open cohort (1984/85–2004/05)
Age group (f) [result not reported] Gender (f) [result not reported] Location (f) [result not reported]
Ng 2010 (Canada)
Multi-variate logistic regression (odds ratio)
Prospective cohort (2000/01 –2002/03)
Age ≥ 65 years (i) [↑] Female (i) [↓] Lower to middle household income (i) [↑] Health utility indexd (i) [↓] Other chronic conditions (i) [↑] Prior hospitalisations (i) [↑] Impact of health problems experienced often or sometimes (i) [↑] Physically inactive (i) [↑] Former or current smoker (i) [↑] Regular alcohol consumption (i) [↓] Current insulin use (i) [↑] ≥ 1 specialist consultations in past 12 months (i) [↑] Residing in high hospital use health region (i) [↑]
Age 12–44 years (ref) Age 45–64 years (i) Male (ref) Highest income (ref) Lower, middle, upper middle (based on quintiles) household income (i) Residence urban or rural (i) No other chronic conditions except diabetes (ref) No prior hospitalisation (ref), Impact of health problems never experienced (ref) Physically active (ref) Moderately active (i) Never smoked (ref) Occasional alcohol consumption (ref) Former or never consumed alcohol (i) Not currently on insulin (ref) Body mass index (i) Daily fruit and vegetable consumption (i) Unmet health care needs (i)
Lin 2010 (Taiwan) [outcome is short-term diabetes ACSC]
Cox proportional hazard regression (relative risk of hospitalisation)
Prospective cohort (1997–2002)
New patient (i) [↓] Age (i) [↓] Age ≥60.5 years (i) [↑]
Existing patients (ref) Age <60.5 years (ref) Number of comorbidities (i) Medium continuity of care (i) Low continuity of care (i) Male (i)
[outcome is long-term diabetes ACSC]
Medium (i) [↑] and low continuity of care (i) [↑] relative to high New patient (i) [↓] Age ≥60.5 years [↑]
High continuity (ref) Age (i) Number of comorbidities (i) Male
Bruni 2009 (Italy)
Multi-level logit model (probability of being hospitalised)
Cross sectional (2003)
Age 65–75 years (i) [↓] Age >75 years (i) [↑] Insulin dependence (i) [↑] Male GP gender (i) [↑]
Age 35–65 years (ref) Gender (i) No insulin (ref) GP female (ref) GP age (i) Practice location rural (i) GP postgraduate qualifications (i) % diabetic patients (i) Endocrinology beds (d)
El-Din 2009 (Saudi Arabia)
Stepwise logistic regression (odds of being hospitalised)
Gender (i) [↑] Presence of nephropathy (i) [↑] HbA1c ≥ 7 mmol/L (i) [↑]
Female (ref) Nephropathy not present (ref) HbA1c <7 mmol/l (ref)
Rizza 2007 (Italy)
Multi-variate logistic regression (odds ratio)
Cross sectional (April–July 2005)
Number hospitalisations previous year (i) [↑]
Education level (i) Length of hospital stay (i) Self-reported health status (i) Sex (i)Age (i)
Gulliford 2004 (England)
Multiple linear regression (chronic hospital admissions per 100 000 persons)
Cross sectional (1999)
% rural patients (ha) [↓] % GPs ≥ 61 years (ha) [↑] % practices with female GP (ha) [↓] % primary care clinic with contraceptive service (ha) [↓]
% patients >75 years (ha) % primary care services with child health surveillance services (ha)
Gulliford 2002 (England)
Deprivation scoree (ha) Per cent households headed by semi or unskilled manual occupation (ha) Per cent with limiting long-term illness (ha)
Percent of households of ethnic minority (ha)
Core findings of the effect of primary care resourcing on diabetes-related ambulatory care sensitive hospitalisations
First author, date
Primary health care resource variable
Direction of association with avoidable hospitalisationsa
GPs per population
Adjusted for facility-level prevalence of diabetes, mental health conditions, heart disease
Patients per GP
No adjustment for clinical health risks
Practice nurses per patient
Adjusted for facility-level unadjusted diabetes prevalence
PHC service availability (service categories: no permanent locally based service, part-time, working hours, 24/7 care)
No adjustment for health status. Adjustment for age, gender and location but not reported
GP contacts in the previous 12 months
Adjusted for individual-level health utility, other chronic conditions, prior hospitalisations, lifestyle behaviours
Diabetes outpatient visits
Adjusted for number of comorbidities and age
Use of diabetes outreach service
No adjustment for health status. Adjusted for age. Outreach service use was a proxy of disease severity.
Patients per GP
Funding incentives to promote better quality care
≥6 PHC clinic visits
Adjusted for presence of nephropathy and HbA1c
Patients per GP
Adjusted for number of hospitalisations in previous year and length of stay and self-reported health status
Gulliford 2004, 2002
GPs per population
No adjustment for health status.
Adjusted for proportion of patients per health authority with a limiting long-term illness. Partnership size is a proxy for better access to multi-disciplinary care team.
Ten studies were identified that met the inclusion criteria and were included in this literature review.
Description of hospitalisation variable
Two measures of hospitalisation were used by all 10 studies (Table 2). The binary outcome; the absence or presence of a diabetes-related hospitalisation during the study period, at patient-level was used by half of the studies [19–22, 41]. Half of the studies measured the rate of hospitalisation by facility or health district area level [17, 18, 42–44]. Lavoie and colleagues  measured the average difference in rates of hospitalisation for chronic conditions between communities with differing levels of PHC service resourcing (reflecting differential access to qualified health personal).
The category of diabetes-related ACSC differed among studies (Table 2), ranging from all diabetes-related ACSC, [18, 20, 22, 41] to chronic ACSC [42–44] and only emergency [17, 19] or acute or non-elective  hospitalisations related to diabetes. Lin and colleagues  analysed hospitalisations for short-term and long-term diabetes-related ACSC separately. Among the ten studies, three studies measured hospitalisation for the same ACSC (i.e. chronic conditions), [42–44] two studies measured diabetes-related hospitalisations [20, 22] and the remaining five studies [17–19, 21, 41] each measured hospitalisation for a different set of ACSC.
Description of primary health care resource variables
Patient use of PHC services measured by the number of PHC clinic visits.
PHC workforce measured by FTE GP and/or nursing staff per facility.
Amount of PHC provided based on the service operating hours (e.g. provision of primary care several days a week or a 24-hour/7-days per week service).
Type of practice, for example, a sole GP service or a general practice group with a multi-disciplinary team.
Payment incentive schemes to improve PHC quality .
Adjusters for confounding
Demographic characteristics of the individual patient or the PHC cohort or the catchment population that potentially use the PHC service.
Health status and/or lifestyle indicators of the individual patient, or the PHC cohort or the catchment population that potentially use the PHC service.
The geographic location of the service, whether rural or urban, as an indicator of access to PHC.
PHC provider characteristics defined by the personal profile of GPs such as age, gender and country of qualification, measured at the individual or facility level.
Social and economic characteristics of individual PHC patients or the PHC service catchment population.
PHC quality was measured by only one study, by the number of quality of care indicators delivered within the evidence based recommended time period. The variables used as potential adjusters for confounding in each of the ten studies are described in Table 3.
Association between hospitalisation and level of primary health care resourcing
With the exception of Dushieko et al.  all studies found significant associations between avoidable diabetes-related hospitalisation and the level of PHC resourcing. These associations were not always inverse (i.e. more PHC resourcing fewer hospitalisations), some were positive (i.e. more PHC resourcing more hospitalisations) (Table 4). A summary of the findings on the association between hospitalisation and the level of PHC resourcing is provided below by category of the PHC resource measure.
PHC GP or nurse workforce FTE per capita
Griffiths et al.  was the only study, that included GPs/capita, to report a contrary finding. This study found that the rate of hospitalisation for non-elective diabetes-related hospitalisations decreased as the number of patients per FTE GP increased. Griffiths also reported the same pattern of a decreased rate of hospitalisation and increased number of patients per FTE with practice nurses. Hospitalisation was measured as the rate of registered patients per facility who experienced one or more diabetes-related hospitalisation .
Number of PHC visits
Three studies [19–21] found the probability of a diabetes-related hospitalisation increased as the number of PHC visits by the patient increased. One study  found that more self-reported visits to the GP in the 12 months previous to a diabetes-related hospitalisation resulted in less subsequent hospitalisation for the same reason.
PHC service availability
A crude measure of primary care service availability based on opening hours, identified that a more resourced service (e.g. PHC service available 24-hour/7-days per week compared to a service available 3 days/week) resulted in less hospitalisation for chronic conditions .
PHC practice size
Gulliford et al.  found that as resources in the primary care service increased, measured by mean partnership size, the hospitalisation rate decreased. Consistent with this was, as the proportion of sole practitioner PHC services within the health authority area increased hospitalisations increased.
PHC payment incentive schemes
Economic incentives for GPs to improve the quality of care to people with diabetes reduced the probability of hospitalisation .
Ten peer reviewed published studies of the association between ambulatory care sensitive diabetes-related hospitalisation and PHC resourcing were located through a rigorous search strategy.
Of the studies found, the measures of hospitalisation were limited to the dichotomous outcome of whether or not an individual experienced an avoidable diabetes-related hospitalisation or any chronic condition ACSC admissions during the study period, or measured the rate of hospitalisations. Other recognised ways of measuring hospitalisation such as cost of hospitalisation was not reported in any of the ten studies.
Even though the reason for hospitalisation was limited to diabetes-related or chronic condition ACSC, studies chose different ways to categorise the group of diagnoses (e.g. short and long-term diabetes-related ACSC) or chose a subset of diagnoses from this group (e.g. hyperglycemic emergency hospitalisations). Some of this variation may be explained by whether PHC access or quality was being evaluated and may reflect the complexity of PHC provision and T2DM disease pathways. For example the hospital outcome measure chosen by Dusheiko and colleagues  was unplanned emergency hospitalisations for short-term diabetes complications - their reason being that improved monitoring (i.e. improved quality of care) of patients with diabetes may increase elective admissions in the short to medium-term and in the longer-term reduce admissions for micro and macrovascular comorbidities .
The PHC resources measured also varied across studies and were used as proxy measures of PHC quality, availability or access. Reviewed studies identified the importance of including predictors of ACSC diabetes-related hospitalisations in the final model so that PHC resourcing was not wrongly attributed to the health outcome. Health status was included in seven of the ten studies (Griffiths et al. adjusted for health status when using standardised diabetes admission ratio as the outcome). Failing to adjust for health status is a weakness, given the known importance of health status for both hospitalisations and use of primary care services.
It is important to consider the complexities and limitations of using ACSC hospitalisations to measure the performance of PHC. By definition, primary health care is the first point of care that is continuous, coordinated and comprehensive whilst being accessible, acceptable and affordable to the population it serves . The role of PHC is diverse and not simply about keeping people out of hospital. Therefore hospitalisation for ACSC can only ever be an incomplete and sometimes poor measure of the performance of PHC. The effect of PHC on ACSC was however the focus of this review, as an interesting policy question.
Much work has been done on rigorous selection of hospitalisations that would most likely be prevented with good ambulatory care [45–48]. Even so, the extent to which PHC can prevent or intervene in disease progression that may result in no or less hospitalisation (e.g. represented by decreased length of stay or a less severe reason for admission) will likely vary across conditions. The implication of this is that the impact of PHC on one ACSC hospitalisation is not uniform for each or across all ACSC hospitalisations. For example, a diagnosis of type 2 diabetes that occurs prior to related impaired kidney function (macroalbuminuria) will provide an opportunity for a comprehensive PHC service to prevent or slow progression to kidney disease. Whereas the same opportunity for PHC to intervene is lost if a diagnosis of diabetes is made, with already established renal impairment. This also highlights the importance of adjusting for individual disease stage  in statistical models.
Limitations of using ACSC hospitalisations to measure the performance of PHC also include those related to the measure of hospitalisation. Variation in hospital admission policies within and between hospitals and decisions made by hospital staff on the need to admit patients are likely to affect rates of hospitalisation for ACSCs . It should also be noted that the quality of and access to PHC may influence some hospital admission policies and staff decisions on patient admission.
In addition, not all possible determinants of hospitalisation for chronic disease related ACSC, some of which were highlighted in the introduction, are accounted for in statistical models. This can distort the estimated impact of PHC on hospitalisation.
With due consideration of these complexities and limitations, ambulatory care sensitive hospitalisations are a useful measure of the performance of PHC at a population level, and of clear interest to policy makers. Hospital administrative data is objective, available and relatively inexpensive to gather. Gradual improvements in the scope and rigour of PHC data collection, that allows for more variables to be included in the models, should improve the accuracy and interpretability of the results of such studies.
The reviewed studies findings were mixed. Seven of the twelve PHC resource variables that had a statistically significant association with ambulatory care sensitive hospitalisations supported the hypothesis that more PHC resources are associated with less hospitalisation for ACSC. However, three of these studies did not adjust for health status [19, 43, 44]. Excluding the results of studies that did not adjust for health status, [19, 43, 44] six PHC resource variables remained and of these three supported the hypothesis that more PHC resources are associated with less hospitalisation for ACSC.
If all 12 significant findings are divided into two categories, by type of primary care variable, a clearer story emerges. Separating out i) the use of primary care services (e.g. n out-patient clinic visits) three of four reported relationships were positive - more visits were associated with higher rates of hospitalisation; from ii) access to primary care (eg GPs/capita, GP/patient list, range of PHC services) or incentives for higher quality of care; six of eight tested relationships (five of six studies) reported a significant inverse association between primary care and hospitalisation; that better access to quality primary care resulted in fewer ACSC hospitalisations. By applying the same categorisation (PHC use or access) to the six studies that adjust for health status, the conclusion remains that better access to primary health care resulted in fewer ACSC hospitalisations.
There is little published evidence on the relationship between PHC resourcing and hospitalisation for diabetes-related ACSC. Studies use a range of measures of primary care and of hospitalisation, creating challenges for interpretation. Also the extent of adjustment for confounders is very mixed with three studies failing to adjust for health status.
While outcomes are mixed, in terms of the direction of the relationship, the impression from this body of work is that access to primary care (as distinct from use – which will be highly confounded by health status) is probably associated with a reduced rate of hospitalisation for diabetes-related ACSC.
Collectively, study findings must still be considered inconclusive, and the relationship between PHC resourcing and hospitalisation for diabetes-related ACSC remains uncertain. Thus additional studies are needed that adjust for a wide range of potential confounders and consider more carefully how best to adjust for disease severity.
Ambulatory care sensitive conditions
Primary health care
This study was supported in part by funding from the Australian National Health and Medical Research Council (ID: 456402).
- World Health Organization, United Nations Children Fund: Report of the international conference on primary health care, Alma-Ata, USSR, 6–12 September 1978 [Internet]. 1978, Geneva: World Health Organisation, [cited 2013 Jan 27]. Available at: http://www.unicef.org/about/history/files/Alma_Ata_conference_1978_report.pdf Google Scholar
- World Health Organisation: The Ottawa charter for health promotion [Internet]. 1986, [cited 2013 Mar 3]. Available at: http://www.who.int/healthpromotion/conferences/previous/ottawa/en/ Google Scholar
- A life course approach to chronic diseases epidemiology. Edited by: Kuh D, Shlomo YB. 2004, USA: Oxford University PressGoogle Scholar
- United Kingdom Prospective Diabetes Study Group: Tight blood pressure control and risk of macrovascular and microvascular complications in type 2 diabetes: UKPDS 38. BMJ. 1998, 317 (7160): 703-713. 10.1136/bmj.317.7160.703.View ArticleGoogle Scholar
- Harris P, Mann L, London J, Phillips P, Webster C: Diabetes Management in General Practice. 2009, Canberra: Diabetes Australia and the Royal Australian College of General Practitioners, 15Google Scholar
- Department of Health and Aging: Chronic disease management (CDM) Medicare items [Internet]. [cited 2013 Jan 10]. Available at: http://www.health.gov.au/internet/main/publishing.nsf/content/mbsprimarycare-chronicdiseasemanagement
- Rutstein D, Berenberg W, Chalmers T, Child C, Fishman A, Perrin E: Measuring the quality of medical care: a clinical method. N Engl J Med. 1976, 294 (11): 582-588. 10.1056/NEJM197603112941104.View ArticlePubMedGoogle Scholar
- Ansari Z: The concept and usefulness of ambulatory care sensitive conditions as indicators of quality and access to primary health care. Aust J Prim Health. 2007, 13 (3): 91-110. 10.1071/PY07043.View ArticleGoogle Scholar
- Billings J, Zeitel L, Lukomnik J, Carey T, Blank A, Newman L: Impact of socio-economic status on hospital use in New York city. Health Affair. 1993, 12 (1): 162-173. 10.1377/hlthaff.12.1.162.View ArticleGoogle Scholar
- Weissman J, Gatsonis C, Epstein A: Rates of avoidable hospitalization by insurance status in Massachusetts and Maryland. JAMA. 1992, 268 (17): 2388-2394. 10.1001/jama.1992.03490170060026.View ArticlePubMedGoogle Scholar
- Sanders J, Solberg B, Gauger M: Breaking barriers to care: a community of solution for chronic disease management. The Journal of the American Board of Family Medicine. 2013, 26 (3): 311-315. 10.3122/jabfm.2013.03.120236.View ArticlePubMedGoogle Scholar
- Bottle A, Millett C, Xie Y, Saxena S, Wachter RM, Majeed A: Quality of primary care and hospital admissions for diabetes mellitus in England. J Amb Care Man. 2008, 31: 226-238. 10.1097/01.JAC.0000324668.83530.6d.View ArticleGoogle Scholar
- Magan P, Alberquilla A, Otero A, Ribera J: Hospitalizations for ambulatory care sensitive conditons and quality of primary care: their relation with socioeconomic and health care variables in the Madrid regional health service (Spain). Med Care. 2011, 49 (1): 17-23. 10.1097/MLR.0b013e3181ef9d13.View ArticlePubMedGoogle Scholar
- Stamp K, Duckett S, Fisher D: Hospital use for potentially preventable conditions in Aboriginal and Torres Strait Islander and other Australian populations. Aust NZ J Public Health. 1998, 22 (6): 673-678. 10.1111/j.1467-842X.1998.tb01468.x.View ArticleGoogle Scholar
- Ansari Z, Barbetti T, Carson NJ, Auckland MJ, Cicuttini F: The Victorian ambulatory care sensitive conditions study: rural and urban perspectives. Soc and Prevent Med. 2003, 48 (1): 33-43. 10.1007/s000380300004.View ArticleGoogle Scholar
- Muenchberger H, Kendall E: Predictors of preventable hospitalization in chronic disease: priorities for change. J Public Health Pol. 2010, 31 (2): 150-163. 10.1057/jphp.2010.3.View ArticleGoogle Scholar
- Dusheiko M, Doran T, Gravelle H, Fullwood C, Roland M: Does higher quality of diabetes management in family practice reduce unplanned hospital admissions. Health Serv Res. 2011, 46 (1p1): 27-46. 10.1111/j.1475-6773.2010.01184.x.View ArticlePubMedPubMed CentralGoogle Scholar
- Griffiths P, Murrells T, Dawoud D, Jones S: Hospital admissions for asthma, diabetes and COPD: is there an association with practice nurse staffing? a cross sectional study using routinely collected data. BMC Health Serv Res. 2010, 10 (1): 276-10.1186/1472-6963-10-276.View ArticlePubMedPubMed CentralGoogle Scholar
- Bruni M, Nobilio L, Ugolini C: Economic incentives in general practice: the impact of pay-for-participation and pay-for-compliance programs on diabetes care. Health Policy. 2009, 90 (2–3): 140-148.View ArticleGoogle Scholar
- El Din M, Al-Mulhim N, Gawwad E: Factors associated with hospitalization for type 2 diabetic patients at a teaching hospital in Saudi Arabia. J Egypt Public Health Assoc. 2009, 84 (1 & 2): 1-19.Google Scholar
- Lin W, Huang I, Wang S, Yang M, Yaung C: Continuity of diabetes care is associated with avoidable hospitalizations: evidence from Taiwan’s national health insurance scheme. Int J Qual Health Care. 2010, 22 (1): 3-8. 10.1093/intqhc/mzp059. 02View ArticlePubMedGoogle Scholar
- Rizza P, Bianco A, Pavia M, Angelillo I: Preventable hospitalization and access to primary health care in an area of Southern Italy. BMC Health Serv Res. 2007, 7 (1): 134-10.1186/1472-6963-7-134.View ArticlePubMedPubMed CentralGoogle Scholar
- Allan D, Cloutier-Fisher D: Health service utilization among older adults in British Columbia: making sense of geography. Can J Aging. 2006, 25 (02): 219-232. 10.1353/cja.2006.0032.View ArticlePubMedGoogle Scholar
- Korenbrot C, Kao C, Crouch J: Funding of tribal health programs linked to lower rates of hospitalization for conditions sensitive to ambulatory care. Med Care. 2009, 47 (1): 88-96. 10.1097/MLR.0b013e3181808bce.View ArticlePubMedGoogle Scholar
- Mitton C, O’Neil D, Simpson L, Hoppins Y, Harus S: Nurse-physician collaborative partnership: a rural model for the chronically ill. Can J Rural Med. 2007, 12 (4): 208-216.PubMedGoogle Scholar
- Shah B, Gunraj N, Hux J: Markers of access to and quality of primary care for Aboriginal people in Ontario Canada. Am J Public Health. 2003, 93 (5): 798-10.2105/AJPH.93.5.798.View ArticlePubMedPubMed CentralGoogle Scholar
- Lindstrom K, Engstrom S, Bengtsson C, Borgquist L: Determinants of hospitalisation rates: does primary health care play a role?. Scand J Prim Health Care. 2003, 21 (1): 15-15. 10.1080/02813430310000500.View ArticlePubMedGoogle Scholar
- Nolan A: An extension in eligibility for free primary care and avoidable hospitalisations:a natural experiment. Soc Sci Med. 2011, 73 (7): 978-985. 10.1016/j.socscimed.2011.06.057.View ArticlePubMedGoogle Scholar
- Pelone F, Kringos D, Valerio L, Romaniello A, Lazzari A, Ricciardi W, Guilio de Belvis A: The measurement of relative efficiency of general practice and the implications for policy makers. Health Policy. 2012, 107 (2–3): 258-268.View ArticlePubMedGoogle Scholar
- Dusheiko M, Gravelle H, Martin S, Rice N, Smith P: Does better disease management in primary care reduce hospital costs? evidence from English primary care. J Health Econ. 2011, 30 (5): 919-932. 10.1016/j.jhealeco.2011.08.001.View ArticlePubMedGoogle Scholar
- Begum N, Donald M, Ozolins I, Dower J: Hospital admissions, emergency department utilisation and patient activation for self-management among people with diabetes. Diabetes Res Clin Pract. 2011, 93 (2): 260-267. 10.1016/j.diabres.2011.05.031.View ArticlePubMedGoogle Scholar
- Tomlin AM, Tilyard MW, Dovey SM, Dawson AG: Hospital admissions in diabetic and non-diabetic patients: a case–control study. Diabetes Res Clin Pract. 2006, 73 (3): 260-267. 10.1016/j.diabres.2006.01.008.View ArticlePubMedGoogle Scholar
- Lakhani A, Coles J, Eayres D, Spence C, Rachet B: Creative use of existing clinical and health outcomes data to assess NHS performance in England: part 1 - performance indicators closely linked to clinical care. BMJ. 2005, 330 (7505): 1426-1431. 10.1136/bmj.330.7505.1426.View ArticlePubMedPubMed CentralGoogle Scholar
- Ansari Z, Laditka J, Laditka S: Access to health care and hospitalization for ambulatory care sensitive conditions. Med Care Res Rev. 2006, 63 (6): 719-741. 10.1177/1077558706293637.View ArticlePubMedGoogle Scholar
- McDermott R, Segal L: Cost impact of improved primary level diabetes care in remote Australian indigenous communities. Aust J Prim Health. 2006, 12 (2): 124-130. 10.1071/PY06031.View ArticleGoogle Scholar
- Mahto R, Venugopal H, Vibhuti V, Mukherjee A, Cherukuri V, Healey B, Baskar V, Buch HN, Singh BM: The effectiveness of a hospital diabetes outreach service in supporting care for acutely admitted patients with diabetes. QJM. 2009, 102 (3): 203-207.View ArticlePubMedGoogle Scholar
- O’Connell J, Wilson C, Manson S, Acton K: The costs of treating American Indian adults with diabetes within the Indian Health Service. Am J Public Health. 2012, 102 (2): 301-308. 10.2105/AJPH.2011.300332.View ArticlePubMedPubMed CentralGoogle Scholar
- Evans NR, Richardson LS, Dhatariya KK, Sampson MJ: Diabetes specialist nurse telemedicine: admissions avoidance, costs and casemix. European Diabetes Nursing. 2012, 9 (1): 17-21. 10.1002/edn.198.View ArticleGoogle Scholar
- Grumbach K: Benefits of primary care and a primary care medical home: a concise summary of the evidence. Centre for Excellence in Primary Care. 2009, University of California San Francisco [Internet], [cited 2012 Dec 12]; Available at: http://familymedicine.medschool.ucsf.edu/cepc/pdf/Benefits%20of%20Primary%20CareMedical%20Home.2.09.pdf Google Scholar
- Freund T, Peters-Klimm F, Rochon J, Mahler C, Gensichen J, Erler A, Beyer M, Baldauf A, Gerlach FM, Szecsenvi J: Primary care practice-based care management for chronically ill patients (PraCMan): study protocol for a cluster randomized controlled trial ISRCTN56104508]. Trials. 2011, 12 (1): 163-10.1186/1745-6215-12-163.View ArticlePubMedPubMed CentralGoogle Scholar
- Ng E, McGrail K, Johnson J: Hospitalization risk in a type 2 diabetes cohort. Health Rep. 2010, 21 (3): 29.PubMedGoogle Scholar
- Gulliford M: Availability of primary care doctors and population health in England: is there an association?. J Public Health Med. 2002, 24 (4): 252-254. 10.1093/pubmed/24.4.252.View ArticlePubMedGoogle Scholar
- Gulliford M, Jack R, Adams G, Ukoumunne O: Availability and structure of primary medical care services and population health and health care indicators in England. BMC Health Serv Res. 2004, 4 (1): 12-10.1186/1472-6963-4-12.View ArticlePubMedPubMed CentralGoogle Scholar
- Lavoie J, Forget E, Prakash T, Dahl M, Martens P, O’Neil J: Have investments in on-reserve health services and initiatives promoting community control improved First Nations’ health in Manitoba. Soc Sci Med. 2010, 71 (4): 717-724. 10.1016/j.socscimed.2010.04.037. 8View ArticlePubMedGoogle Scholar
- Caminal J, Starfield B, Sánchez E, Casanova C, Morales M: The role of primary care in preventing ambulatory care sensitive conditions. The European Journal of Public Health. 2004, 14 (3): 246-251. 10.1093/eurpub/14.3.246.View ArticlePubMedGoogle Scholar
- Fleming ST: Primary care, avoidable hospitalization, and outcomes of care: a literature review and methodological approach. Med Care Res Rev. 1995, 52 (1): 88-108. 10.1177/107755879505200106.View ArticlePubMedGoogle Scholar
- Solberg LI, Peterson KE, Ellis RW, Romness K, Rohrenbach E, Thell T, Smith A, Routier A, Stillmank M, Zak S: The Minnesota project: a focused approach to ambulatory quality assessment inquiry. 1990, 27 (4): 359-367.
- Aiken LH, Lewis CE, Craig J, Mendenhall RC, Blendon RJ, Rogers DE: The contribution of specialists to the delivery of primary care. N Engl J Med. 1979, 300 (24): 1363-1370. 10.1056/NEJM197906143002404.View ArticlePubMedGoogle Scholar
- Gibson O, Segal L, McDermott R: A simple diabetes vascular severity staging instrument and its application to a Torres Strait Islander and Aboriginal adult cohort of north Australia. BMC Health Serv Res. 2012, 12 (1): 185-10.1186/1472-6963-12-185. PubMed PMID: doi:10.1186/1472-6963-12-185View ArticlePubMedPubMed CentralGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1472-6963/13/336/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.