An evaluation of gender equity in different models of primary care practices in Ontario

Background The World Health Organization calls for more work evaluating the effect of health care reforms on gender equity in developed countries. We performed this evaluation in Ontario, Canada where primary care models resulting from reforms co-exist. Methods This cross sectional study of primary care practices uses data collected in 2005-2006. Healthcare service models included in the study consist of fee for service (FFS) based, salaried, and capitation based. We compared the quality of care delivered to women and men in practices of each model. We performed multi-level, multivariate regressions adjusting for patient socio-demographic and economic factors to evaluate vertical equity, and adjusting for these and health factors in evaluating horizontal equity. We measured seven dimensions of health service delivery (e.g. accessibility and continuity) and three dimensions of quality of care using patient surveys (n = 5,361) and chart abstractions (n = 4,108). Results Health service delivery measures were comparable in women and men, with differences ≤ 2.2% in all seven dimensions and in all models. Significant gender differences in the health promotion subjects addressed were observed. Female specific preventive manoeuvres were more likely to be performed than other preventive care. Men attending FFS practices were more likely to receive influenza immunization than women (Adjusted odds ratio: 1.75, 95% confidence intervals (CI) 1.05, 2.92). There was no difference in the other three prevention indicators. FFS practices were also more likely to provide recommended care for chronic diseases to men than women (Adjusted difference of -11.2%, CI -21.7, -0.8). A similar trend was observed in Community Health Centers (CHC). Conclusions The observed differences in the type of health promotion subjects discussed are likely an appropriate response to the differential healthcare needs between genders. Chronic disease care is non equitable in FFS but not in capitation based models. We recommend that efforts to monitor and address gender based differences in the delivery of chronic disease management in primary care be pursued.


Background
Primary care is the foundation of the Canadian health care system. Recent Canadian [1] and international policy recommendations [2] have emphasised the need for investments in primary health care systems to improve efficiencies and reduce inequities. There is convincing evidence that stronger primary health care systems can reduce disparities in health between regions [3]. However few studies have investigated whether the organization of the primary care system impacts on equitable care across individuals.
Evaluations of equity can be seen from two perspectives. Vertical equity addresses whether treatment is preferentially delivered to those with greater health needs, while horizontal equity considers whether there is the provision of equal treatment for equivalent needs [4]. For example, vertical equity would dictate that an individual with multiple health problems should receive greater care than a healthy individual, while horizontal equity would require that two individuals with similar health status receive similar care levels regardless, for example, of their socio-economic status. Both paradigms are important to consider.
Ontario, Canada's largest province, organises primary care practices under different "models of care", most of which emerged following a series of provincial initiatives over the past four decades that aim to build a more accessible, patient oriented system and eliminate the barriers inherent in the traditional Fee For Service (FFS) model [5]. The first attempts at reforming primary care came with the introduction of Community Health Centres (CHC) and Health Service Organizations (HSO) in the 1970s. CHCs are a community orientated model in which providers are salaried. Integral in many CHCs' mission statement are the notions of social justice and equity [6][7][8]. HSO is a capitation based model; a payment structure that offers a fixed monthly remuneration fee based on the age and sex of enrolled patients for basic primary care services, regardless of the number of services provided [9]. A second capitation model which also offered additional accessibility and comprehensiveness incentives, Family Health Networks, (FHNs) was established in the early 2000s. Because compensation in capitation based practices is dissociated from visit number, proponents of this type of remuneration approach expect care to be more equitably dispensed; in response to need with reduced concerns over output. In fact, primary care capitation based funding was recently introduced in New Zealand [10] and Thailand [11] in part in an effort to reduce inequities. Today, capitation based practices and CHCs serve approximately 40%, and 3%, respectively of the population in Ontario.
Some studies have evaluated the impact of these reforms on the quality of the care delivered, [12,13] but none have studied their impact on the equitable delivery of care. In a recent review, the World Health Organization calls for more work evaluating the effect of health care reforms on gender equity [14]. This study evaluates whether gender differences in the primary care experience in each model exist and whether the extent of gender differences between models differs. This study is part of a larger evaluation exploring the impact of primary care reforms on equity.

Design
This study uses data from a cross sectional study conducted in Ontario, Canada in 2005-6; the Comparison of Models in Primary Care (COMP-PC) [15]. Data were gathered from primary care practices, providers (family physicians and nurse practitioners) and patients receiving care at these practices. A detailed description of the overall study methodology is available elsewhere. The study was approved by the Ottawa Hospital Research Ethics Board.

Sample
The COMP-PC study evaluated the performance of FFS, CHC, HSO, and FHN across a number of domains. Table 1 summarizes key features of each model. The study had a recruitment strategy stratified by model. Randomly selected eligible FFS practices (n = 155) and all (n) known and eligible CHC (51), HSO (65), and FHN (94) practices were approached for participation. Recruitment was closed when 35 practices per model agreed to participate or when time constraints didn't permit further recruitment.

Data collection
The study recruited 137 practices, surveyed 5,361 patients in the waiting room sequentially (response rate: 82%) as they presented for their appointment ("index visit") and performed a review of 4,108 randomly/systematically selected charts. Those patients not participating in the survey most frequently sited a lack of time to participate. Surveyed patients (30-50/practice) were required to be under the care of one of the participating providers, aged 18 years or older, not severely ill or cognitively impaired, and able to communicate in English or French either directly or through a translator. Charts reviewed were limited to those of patients ages 18 years and older who had been with the practice at least two years.

Instruments
Surveys were adapted from the Primary Care Assessment Tool (PCAT)-Adult edition [16,17] and supplemented with two additional scales [18,19]. The patient survey was divided into two sections. The first was completed in the waiting room before the encounter with the provider and captured socio-demographic and economic information, and elicited patient's experience on the quality of health service delivery. The second was completed after the appointment with the provider and captured visit-specific information, including a measure of health promotion activity. The survey tool was available in English and French [20]. Translators were used in practices in which a significant proportion of the population was expected to have limited or no English or French language skills.
The chart audit collected patient sex, age, and insurance status and measured preventive care and chronic disease management by comparing chart documentation of these activities against recommended guidelines. We measured the provider's recommendation for a manoeuvre rather than patient compliance, and coded it as "done" if it was performed or recommended/discussed even it not done.

Performance measures
We assessed performance across seven dimensions of health service delivery and three dimensions of technical quality of care ( Table 2). The technical quality of care scales are further described in related manuscripts [ [12,21]

Identify gender differences
We compared the performance scores for women to those of men while adjusting for potentially confounding factors using multi-level multivariate regressions for all evaluations except chronic disease management. For the latter, too few observations per practices were available to warrant adjusting for clustering effect with multilevel analyses. For analyses relying on patient survey data, we adjusted for patient socio-demographic and economic characteristics (identified as SE in Table 3) in one analysis, and added measures of health (identified as H in Table 3) in the second analysis. The analyses including health factors inform the horizontal equity evaluation, while those in which it was omitted inform the vertical equity evaluation. For analyses relying on chart data, we had inadequate information on health status, and so only conducted analyses adjusted for age, rurality, and insurance status. In all analyses, Age*Gender interactions were considered and used where Under capitation remuneration, family physicians received a fixed monthly fee per patient enrolled, independent of the number of visits made to the practice by that patient. The capitation fee is based on the enrolled patient sex and age. FHN physicians receive an additional 10% of the FFS structure for each visit. The later is principally intended to allow for a better monitoring of the services delivered. 3 The base capitation rate is reduced to 50% for patients enrolled to a provider with a practice size exceeding 2,400 4 THAS = Telephone Health Advisory Service -A 24 hrs/7 days a week patient telephone advisory service available to enrolled patients. 5 Each physician is required to provide at least one 3 hour session outside regular hours (evening/week end) per week (up to 5 sessions per group/network/ organization) 6 An incentive bonus that is reduced in relation to the number of visits patients make to non-specialists outside the FHN. 7 A penalty incurred from the capitation fee for visits patients make to non-specialists outside the FHN. 8 Multi-disciplinarity refers to the presence of allied health workers (e.g. dietician, social worker, and pharmacist), excluding nursing staff, but including nurse practitioners. Informed by the Ontario Medical Association's "Comparison of Models" appropriate. Variable imputation was used to avoid casewise deletions. We performed multi-level linear regressions for continuous outcomes using SPSS 16, and multi-level logistic regressions for binary outcomes using the Glimmix procedure in SAS. The analyses were stratified by model. All results shown reflect the effect of being a female compared to being a male.

Compare the extent of gender differences between models
The effect sizes (absolute beta values) of the gender variable in each model derived from the regressions performed to meet objective #1 were compared using the tstatistics to evaluate whether models were significantly different in their gender effect.
When meaningful gender differences are observed, we estimated the adjusted performance level for the "typical" women and men. Using the beta coefficients from the regression equation developed to meet objective #1, we calculated the performance level for the "typical" practice patient.

Characteristics of the study population
The study population was determined to be adequately representative of its underlying population [15]. There were significant differences in several patient characteristics between genders (Table 3). Notably, women surveyed were significantly more likely to report days with poor mental or physical health and limitations related to these conditions. However, self perceived health was similar in both groups.

Health service delivery scales
Differences between genders in all health service delivery measures were not clinically meaningful (≤ 2.2%) in the analyses including and excluding health status variables ( Figure 1).

Technical quality of care scales
Health promotion The odds that at least one health promotion item was discussed at the index visit were lower in women in all models but CHCs ( Figure 2). However, since women have more frequent yearly visits, the overall estimated number of subjects discussed over a 12 months period was not significantly different in the two groups in any model. We observed significant gender differences in the type of subjects discussed at the  [44]. Chronic disease management was assessed against recommended guidelines accepted in Ontario for the management of the conditions [45][46][47][48][49][50][51].
For health promotion, patients were asked to indicate which of 7 subjects were discussed with them on that day's visit. We assessed whether at least one subject was discussed on that visit, and estimated the overall extent of health promotion delivered yearly by multiplying the number of subjects discussed at the index visit by the patient's estimated number of visits to that practice for the year. Preventive care was determined by assessing the performance of 6 indicator manoeuvres in the chart audit. The prevention score was the proportion of preventive manoeuvres for which the individual was eligible that were documented. Finally, chronic disease management was also evaluated by chart audit using 2-4 indicators in each of three conditions (Diabetes, Coronary Artery Disease and Congestive Heart Failure). For each condition the score was derived as for prevention, and the overall chronic disease management score was the average of the individual disease scores. index visit. Women were more likely to have discussed family conflicts in CHCs and FHNs. This effect in was significantly larger than in all other models. Men were significantly more likely to have discussed smoking in FFS and FHNs, and exercise and alcohol consumption in all models but CHCs. The gender effect for exercise discussion was significantly larger in FFS, FHN and HSO compared to CHCs. Figure 3 shows the estimated adjusted likelihood of each subject being discussed in the "typical" women and men in each model. Men were not less likely to report discussing smoking, alcohol, or exercise in CHC than in other models. In contrast, women reported HP discussion for virtually all subjects more frequently in CHCs than other models.
Preventive care The composite prevention score for all 6 manoeuvres in the 3,284 eligible individuals was significantly higher in women than men in all models. The adjusted effect sizes (95% CI) were: CHC 18% (12%, 25%); FFS 21% (15%, 27%); FHN 13% (8%, 19%); HSO 17% (10%, 23%). This was due to the greater adherence to recommended care for the two female specific manoeuvres measured. When these are excluded from the evaluation, the adjusted effect sizes (95% CI) based on the four remaining manoeuvres in the 2,096 patients were: CHC -2% (-10%, 6%); FFS 4% (-3%, 11%); FHN -4% (-11%, 3%); HSO 0% (-6%, 6%). With one exception, there were no significant gender differences in individual manoeuvres (Figure 4). Chronic disease management Overall adherence to recommended guidelines for chronic disease In this column socio-demographic and economic factors used for adjustment in the vertical equity analyses are identified as SE, and health related factors used for adjustment in the horizontal equity analyses are identified as H. LICO = Low Income Cut off, a measure of household deprivation used by Statistics Canada [52].
The following symbols reflect the significance level * p < 0.05, † p < 0.01, ‡ = p < 0.001 compared by Pearson Chi Square or independent t-test. management was significantly inferior in women in FFS (difference of -11.2%, 95% CI: -21.7%, -0.8%), and showed a similar trend in CHCs ( Figure 5). However, there was no statistically significant difference in the gender effect between models. Figure 6 shows the estimated adjusted chronic disease management score in the "typical" women and men in each model. CHCs provided significantly better care to women than other models, while the care received by men was similar for most measures between models. The chronic disease management score in women was not significantly lower in FFS than FHN or HSO. Despite showing a tendency for gender disparity, CHCs were superior to other models in the delivery of chronic disease care for men and women.

Discussion
Women attending FFS practices were significantly less likely to have received chronic disease care according to recommended guidelines. We observed a similar trend in CHCs but not in capitation based practices. We also found differences in the health promotion topics reported being discussed between women and men, and these differences varied by model. Women were more likely than men to report discussing family conflicts in CHCs and FHNs, whereas men were more likely than women to report discussing smoking in FFS and FHN consultation, and discussing exercise and alcohol consumption in all models but CHCs.

Health Service Delivery
Consistent with a previous Canadian report, women reported more frequent visits to their primary care practice than men [22]. However, self reported measures of accessibility as well as other dimensions of health service delivery were not meaningfully different in the two groups. We conclude that the delivery of primary care services is equitable across gender in all models.

Technical quality of care Health promotion
The World Health Organization states that gender equity "... requires that men and women will be treated equally where they have common needs, and that their differences will be addressed in an equitable manner." [23] Men are more likely to smoke and abuse alcohol and illicit drugs than women, [24][25][26] while women are more likely to suffer from family conflicts, [27]  Health promotion across gender -Odds ratio of women relative to men. (Adjusted to socio-economic and health status). 4,794 individuals had provided sufficient information to be included in this analysis. The "HP: At least one subject" variable represents the likelihood that at least one health promotion subject was discussed at the index visit. All other variables represent the likelihood that the subject was discussed at the index visit. Odds ratios are adjusted for patient socio-demographic and economic factors and health status. Statistically significant gender differences (p < 0.05) are indicated by "*". Results of the vertical equity analyses in which health status were not included are consistent with these results.
suggesting that the gender differences observed are likely an appropriate response to the differential healthcare needs between sexes.
We found the smallest gender gap and best performance for women in CHCs. These results may reflect the focus on health promotion and preventive care integral to CHCs, and the substantially longer visits that would allow time for these activities.

Prevention
We observed no significant gender differences in the delivery of colorectal cancer screening and hearing or visual impairment screening in any model but found that men attending FFS were significantly more likely to have been up to date on influenza immunization. Other studies had also found no gender difference in colorectal cancer screening [28] but a higher likelihood of influenza immunization in men receiving care under the Veteran's Health Administration's services, a system that supports both the fee for service and capitation structures [29]. Conclusions about whether gender disparities exist in preventive care is appreciably impacted by the indicators selected. Other studies, as our did, find significantly better preventive scores when conditions specific to women (breast and cervical cancer screening) are included in the overall preventive score, [30] likely because significant investments have been made to promote awareness and compliance for these manoeuvres. Figure 3 Adjusted estimated likelihood of a subject being discussed. (Adjusted for socio-economic and health status). Women were more likely to discuss HP items in CHCs than in any other model. CHCs were statistically superior to all models for all items, except smoking in FHN. Men were usually equally likely to discuss HP items in all models, although men attending FHNs were more likely than those attending HSO to discuss smoking and more likely than those attending FFS to discuss alcohol. The estimated performance for men and women in each model is shown for the "typical" patient; an individual with the most common features: Age 30-49 (except for fall prevention, where it is <75), without a disadvantaged feature (low education, income below low cut off, language barrier, aboriginal status, uninsured), travel distance less than 10 km, not rural, no limitations due to physical or mental health, or problem lasting more than one year, health good-excellent, and the presence of at least one chronic disease. Results of the vertical equity analyses in which health status were not included are consistent with these results. Statistically significant gender differences (p < 0.05) are indicated by "*".

Chronic disease management
Our results suggest that gender gaps in the quality of care received may be dependent on the model of care. Women attending FFS practices but not in capitation based models were significantly less likely to have received recommended care for chronic diseases. Because this study captured the provider's intent for processes of care, the results point to a disparate approach in the primary care providers' management of chronic diseases between men and women in FFS practices rather than, say, gender differences in patient compliance to these processes.
Studies using simulated patients with congestive heart failure found men were more likely undergo clinical investigations [31]. Others have found men to be more likely to receive more evidence based cardiovascular preventive care for aspirin prescription, [32,33] triple antianginal therapy, [34] beta blocker, [35,36] and angiotensin converting enzyme inhibitors [37][38][39]. Evidence for diabetic care is less well documented and doesn't show preferential gender treatment [40,41].
FFS is the most common model of care in Ontario, serving nearly 60% of its population. Critics of the FFS model contend that the "per visit" fee structure encourages shorter, problem focused visits, while capitation or salary based remuneration systems should achieve better care because the provider is not penalized for additional time spent on those with greater needs. The results of this evaluation support this notion.

Impact of primary care reform on gender equity
Our results suggest that primary care reforms have not had a negative impact on the equitable delivery of primary care across gender. In fact, capitation based practices may provide more equitable chronic disease management and influenza immunization than FFS practices.

Strengths and limitations
The survey study population is limited individuals accessing care, and its results cannot be extrapolated to the general population. Because estimates of health service delivery are based on self reported measures, the patient's prior experience and expectation of care, which Because of the small number of events amongst men, the odds ratio confidence interval is unstable. Statistically significant gender differences (p < 0.05) are indicated by "*".

Figure 6
Adjusted estimated likelihood of a subject being discussed -Horizontal equity. The estimated performance for men and women in each model is shown for the "typical patient profile", i.e. an individual with the most common features: Age 70 years or older with public health insurance (rurality "0"). Adherence to recommended guidelines in women was highest in CHC than other models for diabetes and for overall chronic disease management. Adherence to recommended guidelines in men was highest in CHC than other models for chronic disease Figure 5 Overall chronic disease management across gender. 514 patients had at least one of the three indicator chronic diseases and were included in evaluating CDM; 313 had diabetes, and 273 had CAD. Too few patients had CHF (57) to perform a gender evaluation across models. The gender effect is adjusted for age, insurance status, and rurality. Statistically significant gender differences (p < 0.05) are indicated by "*". could plausibly differ by gender, is likely to impact their response.
The evaluation of preventive care and chronic disease management was based on the abstraction of charts. Since these contain very limited patient socio-demographic information we were unable to account for differences in these factors across gender. We also did not capture additional health information to allow us to evaluate whether gender differences in care is related to existing co-morbidities. Finally, we could not evaluate whether patient provider gender concordance is a vehicle to gender disparity. This "within model" approach to evaluating equity has two advantages. First, it eliminates the effect of differences in the profile of the populations within a model for which one could not adjust. It also allows us to evaluate the impact of the primary care reform initiative that addresses remuneration approach on equity.

Conclusions
This is the first study to perform an evaluation of that scope of primary care dimensions. We found the experience of health care service delivery to be similar in women and men. The gender differences that we found in the discussion of healthy lifestyle subjects may be an appropriate and efficient response to prioritizing care in response to differential health needs given limited visit time. This study documents inequities in the delivery of chronic disease care in FFS practices but not in capitation based practices. We recommend that efforts to monitor and address gender based differences in the delivery of chronic disease management in primary care be pursued.