The impact of rural health system reform on hospitalization rates in the Islamic Republic of Iran: an interrupted time series
Arash Rashidian a, Hossein Joudaki a, Elham Khodayari-Moez b, Habib Omranikhoo a, Bijan Geraili c & Mohamad Arab a
a. Department of Health Management and Economics, School of Public Health, Tehran University of Medical Sciences, Poursina Ave, Tehran 1417613191, Islamic Republic of Iran.
b. Faculty of Medicine, Tarbiat Modarres University, Tehran, Islamic Republic of Iran.
c. Aligoodarz Primary Health Care Network, Lorestan University of Medical Sciences, Khorram Abad, Islamic Republic of Iran.
Correspondence to Mohamad Arab (e-mail: email@example.com).
(Submitted: 30 August 2012 – Revised version received: 19 July 2013 – Accepted: 22 July 2013 – Published online: 04 October 2013.)
Bulletin of the World Health Organization 2013;91:942-949. doi: http://dx.doi.org/10.2471/BLT.12.111708
According to the World Health Organization (WHO), the purpose behind universal health coverage (UHC) is to ensure for all people, whether rich or poor, access to effective health services that meet most of their needs without being exposed to the risk of financial hadrship.1,2 In most low- and middle-income countries, these conditions are not met and the effective health services that are available are not accessible to the entire population.1,3
An important prerequisite for UHC is the adequate provision of health services, which depends on factors such as the availability of physicians and hospital beds to meet demand. By adequate service provision we mean the existence of a sufficient number of services of acceptable quality distributed in a manner that allows the population’s health-care needs to be fulfilled. People who lack health insurance coverage or another form of financial protection find it very difficult to access health services, especially inpatient care,4–6 or delay in doing so. This is commonly referred to as the “uninsured access gap”.7–11 When not enough services are covered under a protection plan or when the financial protection for those services that are “covered” is insufficient, households can be deprived of essential health care or face catastrophic costs.12 Delayed care can also lead to hospitalizations that could have been avoided with timely treatment.2,9,13
Poor access to health care and underutilization of health-care services are problems that may be remedied by expanding primary care and social protection schemes through measures such as social insurance or publicly-funded health care.1,3 However, social protection schemes have a mixed effect on hospital utilization – in theory at least.14 By increasing the use of effective outpatient primary care services, improvements in social protection could reduce hospitalizations that are potentially avoidable – i.e. “primary-care-sensitive” hospitalizations – in what is known as an “efficiency effect”. At the same time, people with better social protection are more likely to be hospitalized when they need to be. This would lead to an increase in “referral-sensitive” hospitalizations in what is termed the “access effect”.14,15 The efficiency and access effects of social protection materialize only if health service provision is adequate.
Health-care financing and delivery in rural parts of the Islamic Republic of Iran
Since the mid-1980s, the Islamic Republic of Iran has enjoyed an extensive network of publicly funded primary health-care (PHC) services in rural areas.16 The network has relied primarily on community health workers (behvarz) as first-line service providers. The behvarz are selected from the locality where they live, receive training for two years and provide basic services at a “health house”. A health house is a part of the district PHC network. It is staffed by one or two behvarz who serve a population of about 1500 on average. A health house offers primary services to mothers and children, including vaccinations, prenatal and postnatal care, growth monitoring, family planning, management of common infectious and chronic conditions, and environmental health services. PHC physicians are based in rural health centres and manage the patients referred to them by the behvarz. The PHC network is well organized and is credited with the improvements in health outcomes that have been observed since the 1980s in rural areas.3,17,18
Secondary care services for the Islamic Republic of Iran’s rural population have been less easily accessible. Unlike PHC, which is almost entirely financed by the central government, outpatient care is not free and is delivered mostly by private sector professionals in urban areas. Such care includes, for example, visits to non-PHC physicians, specialist visits, laboratory and radiology services and medicines dispensed at private pharmacies. Access to hospital care for rural residents has been hindered by financial and organizational factors and by the lack of a fully implemented referral system.17 Uninsured rural households used a “rural insurance card” that covered 90% of hospital costs, but only upon referral by a PHC service. The card, which was issued free of charge by the Medical Services Insurance Organization (MSIO), did not cover any outpatient care, apart from the PHC services which were covered by the government funding.
In recent years, concerns have arisen about the PHC’s ability to respond to the needs of the population in light of changing epidemiologic trends, such as the increase in chronic diseases.19 To respond to these changes, in 2005 the Islamic Republic of Iran’s parliament approved a reform that substantially increased the national budget for rural health care.18 The reform covered all villages and towns with populations of up to 20 000 and had two main elements: a family physician programme and a social protection scheme for rural inhabitants (known as “rural insurance”).18 About 6000 physicians and 400 midwives were added to the PHC network in a span of three years as part of the reform.
Since the health system reform process was intended to improve access to more comprehensive preventive and outpatient care, policy-makers expected it to result in a decrease in hospitalizations. Our objective was to assess the immediate and long-term effect of the reform (henceforth referred to as “the intervention”) on hospitalization rates among rural residents, which were the target population.
The intervention was implemented in June 2005. We conducted an interrupted time series study and analysed: (i) monthly hospitalization rates starting two years before the intervention (2003–2007) to assess its immediate effects; and (ii) monthly hospitalization rates up to September 2012 to assess its long-term effects. An interrupted time series study does not require a concurrent “control group” to establish a causal link between an intervention and an outcome.20 However, the use of a comparison group can result in a better understanding of the effects of the intervention on the intervention group.21 For this reason, we used a concurrent series of monthly hospitalization rates for a non-equivalent “comparison group” consisting of primarily urban civil servants and their dependants, who were the target population of another insurance fund also managed by the MSIO.
We hypothesized that the intervention would affect hospitalization rates more strongly in areas where households were poor and hence more likely to benefit from the intervention than affluent households. We also needed to select a province that was not a patient referral hub for other provinces. Lorestan, a province in the western part of the Islamic Republic of Iran, met these criteria (Table 1).
Table 1. Comparative characteristics of the province of Lorestan and of the country as a whole, Islamic Republic of Iran, 2010
The purpose of the intervention was to improve rural households' access to outpatient and hospital care.19,23 All households in villages and small towns were eligible to join the social protection scheme by paying a nominal fee – approximately 0.25 United States dollars – that covered the printing costs of the “insurance” booklet. There was no additional cost for remaining in the scheme. Since most rural household members were employed in the informal sector, the majority of rural households joined the scheme. According to data for the province, 92% coverage had been attained within two years.
The intervention improved access to health care at two levels. First, better pay drew more family physicians and midwives to PHC centres. Between 2004 and 2005 the number of PHC physicians in the province increased from 124 to 191.24 Of the new physicians, many came from the private sector, which they were required to leave to work as a PHC family physician. Others were new graduates or physicians who had previously held managerial posts in the PHC system. Second, households were given access to outpatient services – visits to specialists, radiology services, laboratory and pharmacy services – and to hospital care (located primarily in cities) at a reduced cost. This was facilitated by possession of the “insurance booklet”, signed by the referring family physician. Patients referred by a family physician were expected to pay 10% of inpatient costs and 30% of outpatient costs out of pocket; self-referred patients had no financial coverage.
We obtained population data from the Statistics Centre of the Islamic Republic of Iran and social protection coverage and monthly hospitalization data from MSIO records.
The inpatient services provided by the Imam Khomeini Relief Foundation for poor rural households, which existed before the intervention, could have exerted a confounding effect on our results. However, the Relief Fund had ceased to function within two years of the intervention. We obtained from the Relief Fund its monthly hospitalization claims for rural households for the study period – there were none from 2008 onwards – and included them when calculating the number of hospitalizations. We asked experienced officers to check the data for accuracy. We engaged three physicians and PHC managers in the province to verify the accuracy of all dates and activities.
Sample size and data analysis
We modelled data using segmented regressions to assess causal links between the intervention and the outcome of interest.20,21 It takes at least 48 observations in a segmented regression analysis to be able to assess seasonality in the data.21,25 Our sample included 60 observations (i.e. one per month) – 27 before and 33 after the intervention – to assess the intervention’s immediate effects, and a longer time series – 114 observations – to assess its long-term effects.
To eliminate the effects of population growth, we defined each observation as the number of hospitalizations per month divided by the population under coverage during the year (i.e. the hospitalization rates) (Table 2). We looked for abrupt drops or increases in the hospitalization rate and for gradual changes in trends at the interruption times.
Table 2. Hospitalizations in the intervention group and in the non-equivalent comparison group during the study period, Islamic Republic of Iran, April 2003 to September 2012
We analysed data for the comparison group for the entire study period except for about two years after 2008. This is because during the latter period the MSIO's civil servant fund transferred the care of the elderly population to another insurer and this altered the demographic profile of the population under its coverage.
We conducted several diagnostic assessments. The Durbin-Watson test suggested the existence of autocorrelations, which we corrected using the Cochrane-Orcutt method.26 We estimated the Kolmogorov–Smirnov statistic to check the normality of the residuals and saw that a power transformation correction was required for the analysis of long-term effects. We estimated the Dicky-Fuller statistic to determine if the series was stationary and the Breusch-Pagan statistic to check for heteroscedasticity in the residuals. The results suggested a normal residual distribution and a lack of seasonality and heteroscedasticity.26
The first “interruption” time was set at June 2005, when the intervention was first implemented. To identify the next interruption point, we used segmented regressions with multiple time periods for each point and monitored the time series to find the first point showing a significant change in the trend (Appendix A, available at: https://docs.google.com/file/d/0B32jQZp_KmJzeHFmRUhwRU5ySGs/edit?usp=sharing). All analyses were conducted using R statistical software version 2.12.1 (The R Foundation for Statistical Computing, Vienna, Austria).
Over 700 000 inhabitants of rural areas, including small towns, were covered by the intervention, and about 150 000 urban inhabitants were covered by the MSIO fund for civil servants (Table 2). From 2003 to 2007, the annual hospitalization rate in the population covered by the intervention increased from 44.3 to 65.6 per 1000 inhabitants. The corresponding rates in the population covered by the MSIO's fund for civil servants were 92.7 in 2003 and 95.7 in 2007. In 2011, the annual hospitalization rate for the intervention group and the comparison group was 62.5 and 78.8 per 1000 inhabitants, respectively.
Immediate effect on hospitalization rate
No significant increase or decline in the hospitalization rate was noted in either the intervention or the comparison group before the intervention. The pre-intervention rates remained stable in both groups (Table 3) even though the baseline monthly hospitalization rate per 1000 inhabitants was markedly lower in the intervention group than in the comparison group.
Table 3. Estimated coefficients of segmented regression model for hospitalization rates in the intervention group and the non-equivalent comparison group before and after the intervention, Islamic Republic of Iran, April 2003 to March 2008
We looked for changes in slope and intercept in the comparison group to check for confounding factors unrelated to the intervention – e.g. disease epidemics, data recording policies at the MSIO – that might have occurred at the same time as the intervention and biased the findings. We observed no significant changes in slope or intercept in the hospitalization rate of the comparison group after the intervention (P > 0.4). However, we did note a significant change in the regression slope (P < 0.001) in the intervention group right after the intervention started (Table 3). This finding suggests that the intervention resulted in a greater likelihood that people would use hospital services (Fig. 1). To develop the model for the hospitalization rate in the intervention group, we transformed the model back to its original coefficients (after correcting for autocorrelation; Yt − ρYt-1). Hence, for Model 1:Fig. 1Segmented regression model showing hospitalization rates over the short-term in the intervention group, April 2003 to March 2008 Note: The intervention was implemented in June 2005. where Y is the hospitalization rate, ρ is the autocorrelation parameter and equals 0.125516; T is pre-intervention slope; P is the change in slope and D is the change in intercept.
Fig. 1. Segmented regression model showing hospitalization rates over the short-term in the intervention group, April 2003 to March 2008
According to Model 1, the hospitalization rate in the intervention group increased, on average, by about 4.6 hospitalizations per 100 000 insured people per month. Fig. 1 shows that before the intervention the slope was close to zero but showed a much steeper forward incline after the intervention. Segmented regression analysis conducted in both groups revealed that the post-intervention slopes in the intervention group and the comparison group differed statistically (P = 0.004). After the intervention, the intervention group had about 7 monthly hospitalizations more per 100 000 insured people than the comparison group (Appendix B, available at: https://docs.google.com/file/d/0B32jQZp_KmJzN3YtWWp1cXRLbWc/edit?usp=sharing).
Long-term effect on hospitalization rate
In the intervention group we observed a second statistically significant “interruption” point in the trend in hospitalization rate 40 months after the start of the intervention (Fig. 2). At that point the upward trend in the hospitalization declined significantly and returning to the modest upward trend that had existed before the intervention started (P < 0.001). The slopes before the intervention and after the second interruption point were not statistically different (P = 0.75) (Appendix C, available at: https://docs.google.com/file/d/0B32jQZp_KmJzWUdjS1pGcGpNNzA/edit?usp=sharing). As a result, the long-term effect of the intervention can be described as an abrupt increase in the hospitalization rate lasting more than one year, followed by a secular upward trend that was comparable to the trend that had existed before the intervention (Fig. 2). Analysis of the comparison group over the long term showed an unexplained significant reduction in the hospitalization rate – seen as a change in the intercept – after 2009 (P < 0.001).
Fig. 2. Segmented regression model showing hospitalization rates over the long term in the intervention group, April 2003 to September 2012
A family physician programme and a social protection scheme for rural inhabitants in the Islamic Republic of Iran resulted in a modest and statistically significant increase in the hospitalization rate. The upward trend in the rate remained relatively constant for about one year and then began to decline, perhaps under the influence of external factors not observed by our study, since we also noted a reduction in the hospitalization rate in the comparison group at about the same time.
The intervention increased the utilization of hospital beds in a population that had historically underutilized hospital services. This suggests that the “access effect” of the intervention outweighed its potential “efficiency effect”. It appears, therefore, that while family physicians provided rural inhabitants with a “point of referral” to specialists, the social protection scheme increased their actual use of specialists, and, ultimately, of hospital services.
Our results differ from those of previous studies that claim that expanding primary care services – or access to such services – would result in a decrease in the hospitalization rate by reducing primary-care-sensitive hospitalizations.14,27–32 All such studies but one32 were conducted in high-income countries with high baseline access to inpatient hospital services and have one major limitation: they focused exclusively on primary-care-sensitive avoidable hospitalizations. Only Friedman & Basu (2001) measured total hospitalizations and observed a reduction in all hospitalizations among children in the state of New York, United States of America.31
Other studies have led to conclusions similar to those derived from ours. In a study in Canada, improving access to primary care resulted in a reduction in primary-care-sensitive hospitalizations among the elderly but did not change total hospitalization rates.33 In Ireland, Nolan et al. (2011) observed no changes in avoidable hospitalizations as a result of improved access to primary care.34 A study conducted in the United States revealed that reductions in the hospitalization rate were not linked to the use of primary care services.35
Saha et al. (2007) had hypothesized, like we did, that improving access to primary care – by expanding the Medicaid programme in the state of Oregon in the United States to cover an additional 100 000 low-income families – would result in fewer hospitalizations. Instead, they observed an increase in the hospitalization rate.36 In their study, the baseline hospitalization rate in the intervention group was lower than in the rest of the population. In ours, the hospitalization rate at baseline was also substantially lower in the rural than in the urban population. These findings suggest that the increase in the hospitalization rate that we observed in the rural population reflects a pre-existing unmet need for hospitalizations. We would expect the increase to have occurred in referral-sensitive hospitalizations,37 although this should be the focus of future studies.
Our study has important methodological strengths. We took advantage of the opportunity to perform a natural experiment and used robust research methods.19,21 We made certain that all statistical assumptions were met. We also looked for potential confounding influences during the intervention period that might have biased the results and found none. The study involved a relatively large number of observations (months) that resulted in enough power to test the hypothesis and obtain significant results at the desired confidence level.
The study also has important limitations. It was conducted in only one province, whereas ideally it should have been conducted in a nationally representative sample. On the other hand, by focusing on a single province we were able to better monitor data quality and contextual factors that could have influenced hospitalizations in the intervention group. We lacked data on hospitalization length and causes of admission. Future studies should explore whether hospitalization patterns changed after the rural health system reform.
Our study has important policy implications. If the hospitalization rate in the intervention group had continued to increase for over one year after the intervention as a response to unmet needs, it should have continued to rise until the unmet need was satisfied – i.e. until it had reached a level similar to that in the comparison group. However, the hospitalization rate in the intervention group did not reach the level observed in the comparison group. This is difficult to explain because the two groups were not comparable in terms of demographic or access characteristics. The comparison group was predominantly composed of more affluent urban residents with better geographical access to specialist and hospital care than the intervention group.
To conclude, programmes that focus on improving access to primary care do not necessarily lead to reduced hospitalization rates and hence to lower hospitalization-related expenditures. Improved accessibility offsets potential efficiency gains from the intervention, especially in societies with an unmet need for hospitalization.
We thank the head of research at the Tehran University of Medical Sciences for funding the study. We also thank T Nazari and M Toolabi at the MSIO in Lorestan and the Imam Khomeini Relief Foundation, and H Barkhordari, E Jamshidbeigi and A Khosravi at the Provincial and National Primary Health Care Network Management Units for their invaluable assistance.
The Tehran University of Medical Sciences funded the study (Contract No. 9026; April 2009).
- The world health report – health systems financing: the path to universal coverage. Geneva: World Health Organization; 2010. Available from: http://www.who.int/whr/2010/en/ [accessed 3 September 2013].
- Carrin G, James C, Evans DB. Achieving universal health coverage: developing the health financing system. Geneva: World Health Organization; 2005. Available from: http://www.who.int/health_financing/documents/cov-pb_e_05_1-universal_cov/en/index.html [accessed 3 September 2013].
- The world health report 2008: primary health care – now more than ever. Geneva: World Health Organization; 2008. Available from: http://www.who.int/whr/2008/en/ [accessed 3 September 2013].
- Saksena P, Antunes AF, Xu K, Musango L, Carrin G. Mutual health insurance in Rwanda: evidence on access to care and financial risk protection. Health Policy 2011; 99: 203-9 http://dx.doi.org/10.1016/j.healthpol.2010.09.009 pmid: 20965602.
- Cheng SH, Chiang TL. The effect of universal health insurance on health care utilization in Taiwan: results from a natural experiment. JAMA 1997; 278: 89-93 http://dx.doi.org/10.1001/jama.278.2.89 pmid: 9214512.
- Sepehri A, Simpson W, Sarma S. The influence of health insurance on hospital admission and length of stay – the case of Vietnam. Soc Sci Med 2006; 63: 1757-70 http://dx.doi.org/10.1016/j.socscimed.2006.04.029 pmid: 16766108.
- Kataoka SH, Zhang L, Wells KB. Unmet need for mental health care among U.S. children: variation by ethnicity and insurance status. Am J Psychiatry 2002; 159: 1548-55 http://dx.doi.org/10.1176/appi.ajp.159.9.1548 pmid: 12202276.
- Ayanian JZ, Weissman JS, Schneider EC, Ginsburg JA, Zaslavsky AM. Unmet health needs of uninsured adults in the United States. JAMA 2000; 284: 2061-9 http://dx.doi.org/10.1001/jama.284.16.2061 pmid: 11042754.
- Cummings JR, Lavarreda SA, Rice T, Brown ER. The effects of varying periods of uninsurance on children’s access to health care. Pediatrics 2009; 123: e411-8 http://dx.doi.org/10.1542/peds.2008-1874 pmid: 19254977.
- Long SH, Marquis MS. The uninsured 'access gap' and the cost of universal coverage. Health Aff 1994; 13: 211-20 http://dx.doi.org/10.1377/hlthaff.13.2.211.
- Cunningham PJ, Hadley J. Differences between symptom-specific and general survey questions of unmet need in measuring insurance and racial/ethnic disparities in access to care. Med Care 2007; 45: 842-50 http://dx.doi.org/10.1097/MLR.0b013e318053678f pmid: 17712254.
- Kavosi Z, Rashidian A, Pourreza A, Majdzadeh R, Pourmalek F, Hosseinpour AR, et al., et al. Inequality in household catastrophic health care expenditure in a low-income society of Iran. Health Policy Plan 2012; 27: 1-7 http://dx.doi.org/10.1093/heapol/czs001 pmid: 21324972.
- Weissman JS, Gatsonis C, Epstein AM. Rates of avoidable hospitalization by insurance status in Massachusetts and Maryland. JAMA 1992; 268: 2388-94 http://dx.doi.org/10.1001/jama.1992.03490170060026 pmid: 1404795.
- Dafny L, Gruber J. Public insurance and child hospitalizations: access and efficiency effects. J Public Econ 2005; 89: 109-29 http://dx.doi.org/10.1016/j.jpubeco.2003.05.004.
- Gusmano MK, Rodwin VG, Weisz D. A new way to compare health systems: avoidable hospital conditions in Manhattan and Paris. Health Aff (Millwood) 2006; 25: 510-20 http://dx.doi.org/10.1377/hlthaff.25.2.510 pmid: 16522605.
- Kruk ME, Porignon D, Rockers PC, Van Lerberghe W. The contribution of primary care to health and health systems in low- and middle-income countries: a critical review of major primary care initiatives. Soc Sci Med 2010; 70: 904-11 http://dx.doi.org/10.1016/j.socscimed.2009.11.025 pmid: 20089341.
- Islamic Republic of Iran: health sector review. Washington: The World Bank Group Human Development Sector, Middle East and North Africa; 2007 (Report No: 39970). Available from: http://www.worldbank.org/en/topic/health/research/all?docty_exact=Health+Sector+Review&qterm=&lang_exact=English&count_exact=Iran%2C+Islamic+Republic+of [accessed 3 September 2013].
- Takian A, Rashidian A, Kabir MJ. Expediency and coincidence in re-engineering a health system: an interpretive approach to formation of family medicine in Iran. Health Policy Plan 2011; 26: 163-73 http://dx.doi.org/10.1093/heapol/czq036 pmid: 20732859.
- Takian AH, Doshmangir L, Rashidian A. Implementing family physician programme in rural Iran: exploring the role of an existing primary health care network. Fam Pract 2013; 30: 551-9 http://dx.doi.org/10.1093/fampra/cmt025 pmid: 23788202.
- Cook TD, Campbell DT. Quasi-experimentation: design and analysis issues for field settings. Boston: Houghton Mifflin; 1979.
- Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D. Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther 2002; 27: 299-309 http://dx.doi.org/10.1046/j.1365-2710.2002.00430.x pmid: 12174032.
- Health Technology Assessment Office. Medical Services Insurance Organization's performance report. Tehran: Medical Services Insurance Organization; 2010. Persian.
- National Unit for PHC Expansion and Health Promotion. The operational instructions for family medicine and behvarz. 8th ed. Tehran: Ministry of Health and Medical Education; 2007. Persian.
- Center for PHC Network Development and Health Promotion. Report on the family physician and rural insurance program operational conduct. Tehran: Ministry of Health and Medical Education; 2007.
- Time series analysis. In: Pharmacoepidemiology: behavioural and cultural themes. Newcastle: Center for Clinical Epidemiology and Biostatistics; 2001.
- Kutner MH, Nachtsheim CJ, Neter J. Applied linear regression models. 4th ed. Chicago: Irwin/McGraw-Hill; 2004.
- Parchman ML, Culler S. Primary care physicians and avoidable hospitalizations. J Fam Pract 1994; 39: 123-8 pmid: 8057062.
- Ionescu-Ittu R, Jane McCusker J, Ciampi A, Vadeboncoeur AM, Danièle Roberge D, Larouche D, et al., et al. Continuity of primary care and emergency department utilization among elderly people. Canadian Medical Association 2007; 177: 1362-8.
- Lindström K, Engström S, Bengtsson C, Borgquist L. Determinants of hospitalisation rates: does primary health care play a role? Scand J Prim Health Care 2003; 21: 15-20 http://dx.doi.org/10.1080/02813430310000500 pmid: 12718455.
- Laditka JN, Laditka SB, Probst JC. More may be better: evidence of a negative relationship between physician supply and hospitalization for ambulatory care sensitive conditions. Health Serv Res 2005; 40: 1148-66 http://dx.doi.org/10.1111/j.1475-6773.2005.00403.x pmid: 16033497.
- Friedman B, Basu J. Health insurance, primary care, and preventable hospitalization of children in a large state. Am J Manag Care 2001; 7: 473-81 pmid: 11388127.
- Guanais F, Macinko J. Primary care and avoidable hospitalizations: evidence from Brazil. J Ambul Care Manage 2009; 32: 115-22 http://dx.doi.org/10.1097/JAC.0b013e31819942e51 pmid: 19305223.
- Menec VH, Sirski M, Attawar D, Katz A. Does continuity of care with a family physician reduce hospitalizations among older adults? J Health Serv Res Policy 2006; 11: 196-201 http://dx.doi.org/10.1258/135581906778476562 pmid: 17018192.
- Nolan A. An extension in eligibility for free primary care and avoidable hospitalisations: a natural experiment. Soc Sci Med 2011; 73: 978-85 http://dx.doi.org/10.1016/j.socscimed.2011.06.057 pmid: 21831496.
- Ricketts TC, Randolph R, Howard HA, Pathman D, Carey T. Hospitalization rates as indicators of access to primary care. Health Place 2001; 7: 27-38 http://dx.doi.org/10.1016/S1353-8292(00)00035-6 pmid: 11165153.
- Saha S, Solotaroff R, Oster A, Bindman AB. Are preventable hospitalizations sensitive to changes in access to primary care? The case of the Oregon Health Plan. Med Care 2007; 45: 712-9 http://dx.doi.org/10.1097/MLR.0b013e318053717c pmid: 17667304.
- Shah BR, Gunraj N, Hux JE. Markers of access to and quality of primary care for aboriginal people in Ontario, Canada. Am J Public Health 2003; 93: 798-802 http://dx.doi.org/10.2105/AJPH.93.5.798 pmid: 12721147.