Drinking Patterns, Gender and Health III: Avoiding vs. Seeking Healthcare

Abstract

Background

Inability to predict most health services use and costs using demographics and health status suggests that other factors affect use, including attitudes and practices that influence health and willingness to seek care. Alcohol consumption has generated interest because heavy, chronic consumption causes adverse health consequences, acute consumption increases injury, and moderate drinking is linked to better health while hazardous drinking and alcohol-related problems are stigmatized and may affect willingness to seek care.

Methods

A stratified random sample of health-plan members completed a mail survey, yielding 7884 respondents (2995 male/4889 female). We linked survey data to 24 months of health-plan records to examine relationships between alcohol use, gender, health-related attitudes, practices, health, and service use. In-depth interviews with a stratified 150-respondent subsample explored individuals’ reasons for seeking or avoiding care.

Results

Quantitative results suggest health-related practices and attitudes predict subsequent service use. Consistent predictors of care were having quit drinking, current at-risk consumption, cigarette smoking, higher BMI, disliking visiting doctors, and strong religious/spiritual beliefs. Qualitative analyses suggest embarrassment and shame are strong motivators for avoiding care.

Conclusions

Although models included numerous health, functional status, attitudinal and behavioral predictors, variance explained was similar to previous reports, suggesting more complex relationships than expected. Qualitative analyses suggest several potential predictive factors not typically measured in service-use studies: embarrassment and shame, fear, faith that the body will heal, expectations about likelihood of becoming seriously ill, disliking the care process, the need to understand health problems, and the effects of self-assessments of health-related functional limitations.

INTRODUCTION

In recent years, various researchers have attempted to predict health care service use and costs as part of efforts to improve methods for risk-adjustment and capitation, to develop information-based systems for assessing quality of care, and to assess the health of populations (Majeed et al., 2001). Most such models use diagnostic or claims data and basic demographic information as predictors, although some models have added survey-based information assessing self-reported health and functional status, chronic conditions, and occasionally, health-related behaviors (FitzHenry & Shultz, 2000; Rice & Smith, 2001). In general, these models explain only a small fraction of the variance in use or costs of services—generally ranging from lows of 3–15% (most frequently) to highs of 20–30% and (rarely) 40–50% or more (FitzHenry & Shultz, 2000).

This inability to explain most of what produces health care use and costs has focused analyses on examining, in greater detail, factors that might be more useful predictors. A common model used to frame these analyses is the Andersen behavioral model of healthcare utilization (Andersen & Newman, 1973; Andersen, 1968; Andersen, 1995b), which includes three broad categories of factors: predisposing, enabling, and need factors. Most risk-adjustment models focus on need (e.g., health and functional status) and basic predisposing factors (e.g., age and gender) (FitzHenry & Shultz, 2000), while some models have examined other factors, particularly enabling factors (e.g., access to care, copayments). Taken together, this work suggests that enabling and medical need factors are among the most important determinants of health care utilization, particularly once the important predisposing factors of age and sex have been taken into account (Anderson & Bartkus, 1973; Berki & Ashcraft, 1979; Evashwick et al., 1984; Muller, 1986; Mutran & Ferraro, 1988; Riley et al., 1993; Verbrugge & Patrick, 1995; Wolinsky et al., 1998). Yet, to date, more comprehensive models have not substantially improved variance explained (c.f., Green & Pope, 1999).

It is in this context that researchers have begun to examine other factors that might affect use of health care services, particularly attitudes and practices that might affect both health and peoples’ willingness to seek health care. Of these, alcohol consumption has generated particular interest because 1) epidemiological evidence shows that heavy, chronic alcohol consumption leads to multiple adverse health consequences, including increased mortality (need factors) (Anderson et al., 1993; US Department of Health and Human Services et al., 1997), 2) heavy, acute alcohol consumption increases the risk of injury (need factors) (Anda et al., 1988; Cherpitel et al., 2003; Murgraff et al., 1999; Vinson et al., 2003), 3) moderate drinking has been linked to better health and health-related outcomes (need factors) (Dufour, 1994; Gunzerath et al., 2004; US Department of Health and Human Services et al., 2003), 4) hazardous drinking and alcohol-related problems are stigmatized behaviors and conditions that might affect individuals’ willingness to seek care (predisposing factors), and 5) drinking practices tend to cluster with other health-related attitudes and practices that might affect both health (e.g., healthier lifestyles [Polen et al., companion paper #1; Slater et al., 1999] and willingness to seek care (e.g., smoking, risk taking—need and predisposing factors).

Efforts to understand these relationships have produced unexpected and conflicting findings. Although alcohol-related health consequences suggest that health care use should be significantly higher among hazardous or harmful drinkers and lower among moderate drinkers and abstainers, examinations of the effects of alcohol consumption on use of health care services generally fail to document increased utilization or costs among heavier drinkers. In fact, most studies report lower outpatient service use among heavier drinkers compared to light or moderate drinkers, and find more service use among former drinkers and lifelong abstainers than among heavier drinkers (Armstrong et al., 1998; Kunz, 1997; Manning et al., 1991; Polen et al., 2001; Rice & Duncan, 1995; Rice et al., 2000; Zarkin et al., 2004). Some investigators (Armstrong et al., 1998) speculate that such findings might be due to: (1) inattention by heavier drinkers to health problems, (2) spurious increases in service use among former drinkers resulting from “sick quitters,” whose illnesses resulted from heavy alcohol consumption, or (3) lower rates of illness among moderate drinkers. Evidence supports these contentions. Rice and colleagues (Rice et al., 2000) showed that former heavy drinkers (five or more drinks/day at least weekly for two or more months) had significantly higher rates of outpatient visits, and higher odds of hospitalization than either abstainers or current drinkers. They suggest that these results are consistent with the idea that heavy drinkers avoid preventive or usual care, become ill, quit drinking, and then consume relatively large amounts of health care services. Additional evidence supporting this “sick quitter” hypothesis, particularly among women, comes from studies showing cessation of drinking among those who reported declines in health (Baumeister et al., 2006; Eigenbrodt et al., 2000), and studies that predict abstinence as a function of health status (Green et al., 2003). In our companion paper (Green et al., 2010, companion paper #2) we found that patterns of alcohol consumption had only limited effects on preventive service use, although individuals who reported hazardous or harmful alcohol use were less likely to receive some forms of preventive services.

Here, we address questions arising from these varied findings and hypotheses, using a mixed-methods approach and Andersen’s (1995) model of health service use as our conceptual framework. We present data from a survey of a health plan’s adult membership combined with health plan records of medical utilization and in-depth interviews of a subsample of survey respondents, examining the “sick quitter” hypothesis and testing whether previous relationships between alcohol consumption and service use result from confounding with health related attitudes and practices. In quantitative analyses, we control for need factors (health and functional status) and examine the relationships between predisposing health-related behaviors and attitudes, particularly drinking status, enabling factors such as copayments and having a primary care provider, and service use, at the same time evaluating differences in these relationships among men and women. Using qualitative analyses, we examine whether people quit drinking alcohol as a result of poor health and their reasons for avoiding or seeking health care, including how patterns of alcohol consumption affect these decisions.

METHODS

Setting

The study setting is Kaiser Permanente Northwest, a not-for-profit prepaid group practice health plan that provides outpatient and inpatient medical, mental health, and addiction services to about 480,000 people in northwest Oregon and southwest Washington State, USA. Polen and colleagues’ companion paper #1 includes further details.

Sample and survey

Study participants were 2995 male and 4889 female (7,884 total) HMO members aged 18–64 who responded to a Health & Health Practices survey conducted from October 2002 through mid-April 2003. Surveys were sent to 15,000 members (8500 women/6500 men) who had at least 12 months of health plan membership prior to sample extraction. Details on survey design, sampling methods, and characteristics of sample respondents and non-respondents are included in Polen and colleagues’ companion paper #1.

Study measures

Details on measures of socio-demographic characteristics, health-related attitudes and values, and health-related practices are described in companion paper #1. Alcohol consumption measures were identical to those described in companion paper #2, details on barriers to and facilitators of care seeking, and health and functional status are described also described in Green and colleagues’ (companion paper #2). Quantitative measures included in analyses for this paper, but not included in companion papers #1 or #2, are described below.

Health care service use

We computed the following measures of health care service use for the 12 months following each respondent’s survey return: number of visits to urgent care, number of visits to the emergency department, number of primary care visits, number of specialty outpatient visits (except mental health and addiction medicine), number of telephone calls, and number of inpatient days. We used binary indicators for having had specialty mental health visits, specialty addiction medicine visits, and self-reported service use not paid for by the health plan.

Quantitative analyses

We explored the relationship between the independent variables of interest and measures of health care service use in two steps. First, independent variables were grouped according to whether they were predisposing factors, enabling factors, or need factors. Predisposing factors were further broken down into three categories: demographic characteristics, health-related attitudes, and health-related behaviors. We then regressed each indicator of health care service use on the predictors that comprised each block (in separate regression equations), with a result of five regression analyses for each outcome. Every analysis controlled for months of health plan eligibility, and included gender, alcohol use, and gender by alcohol use interactions as predictors of later service use. Separate analyses were computed for drinking status and, among current drinkers, whether individuals were AUDIT positive.

For each regression analysis, we noted independent variables that were significantly related to a particular outcome at alpha ≤ .05. Significant predictors for each outcome variable from each block were then simultaneously entered into one regression equation (non-significant variables were dropped) to determine which variables remained uniquely predictive. For each outcome variable, if the interaction of gender and alcohol use was not significant, we performed a regression analysis without the interaction term and report those results. We evaluated all tests of statistical significance using robust estimates of the standard error.

Each outcome variable, with the exception of specialty mental health visits and specialty outpatient visits, was a count-based measure of service use. Specialty mental health visits and other specialty outpatient visits were dichotomized indicating whether or not a visit was made (as there was little variation in the visit counts of these variables). Thus, the relationship between predictors and binary outcomes was evaluated using logistic regression analyses. All of the other outcome variables showed overdispersion and did not fit a Poisson distribution well, so were evaluated with negative binomial regression. For logistic regression analyses, we report odds ratios and confidence intervals for each significant independent variable. For negative binomial regression analyses, we report incidence rate ratios and confidence intervals for each significant predictor.

Qualitative interview sampling and recruitment

We interviewed 150 survey respondents, selected to represent various drinking statuses and health care usage patterns, and to represent women and men equally, increasing heterogeneity in the sample across these three dimensions. Deliberate sampling for heterogeneity is useful when seeking a full range of responses and opinions (Blankertz, 1998).

From April 28, 2003, through November 25, 2003, we mailed recruitment letters to 316 of 4477 survey respondents who agreed to be contacted for an interview. Letters provided information about the study and offered a $50 gift card to a local one-stop shopping center chain for participating in a 1-hour interview. We were unable to contact 85 individuals (27%) of the 316 participants to whom we mailed letters. Of those contacted, 81 of 231 refused to complete the interview. Thus, among contacted individuals the participation rate was 65%; overall participation was 52.5%. In total, 150 participants gave informed consent for the study and completed interviews.

Interview participants

Interview participants (75 women, 75 men) ranged in age from 21 to 64 years, with an average age of 46.2 and standard deviation of 12.4 years. Most participants were white (90%), consistent with both the health plan membership and surrounding geographic area; 2% reported mixed racial heritage, 3.3% indicated that they were black or African American, 0.07% indicated that they were Asian/Pacific Islanders, and 0.07% indicated being of American Indian/Alaska Native heritage. Across reported races, 1.3% reported Hispanic ethnicity.

We defined six mutually exclusive drinking status categories based on at-risk drinking guidelines (Gunzerath et al., 2004) and respondents’ survey responses, and used these categories to select interview participants. These included: 1) Lifelong abstainers (persons who had consumed fewer than 12 drinks in their lifetime); 2) Former drinkers (persons indicating that they used to drink, but no longer drink); 3) Low-risk drinkers (current drinkers who scored less than 8 on the AUDIT; averaged 1 drink or fewer per day for women and 2 drinks per day or fewer for men; and if they drank heavily at times [4+ drinks per occasion for women, 5+ for men], it was less than monthly); 4) Hazardous/harmful drinkers (individuals who scored 8 or above on the AUDIT; 5) Immoderate drinkers (for women, average consumption > 1 drink/day; for men, average consumption > 2 drinks/day; but scored below 8 on the AUDIT; and 6) Episodic heavy drinkers (response of “monthly” or “greater than monthly” on the frequency of heavy drinking days question [5+ drinks per occasion for men and 4+ drinks per occasion for women], but scored below 8 on the AUDIT and did not qualify in the category of immoderate drinkers).

Participants were also selected to represent four health care use categories, defined based on health plan records of outpatient service use in the year prior to the date the survey sample was identified. These categories included: 1) No visits (zero visits in the year prior), 2) Low use (1–9 visits in the year prior), 3) Moderate use (10–15 visits in the year prior), and 4) High use (16+ visits in the year prior).

The combination of the six drinking variables and four service-use variables created 24 potential pools from which we hoped to recruit six participants each, balanced on gender (3 men, 3 women). When individuals in particular pools were exhausted, we sampled from the closest available pool (based on service use or drinking category, as necessary). Among men, we interviewed 11 lifelong abstainers, 16 former drinkers, 12 low-risk drinkers, 15 hazardous/harmful drinkers, 9 immoderate drinkers, and 12 episodic heavy drinkers. Among women, we interviewed 10 lifelong abstainers, 15 former drinkers, 10 low-risk drinkers, 19 hazardous/harmful drinkers, 19 immoderate drinkers, and 3 episodic heavy drinkers. Eighteen male interviewees had no health service use in the prior 12 months, 27 had low use, 18 had moderate use, and 12 had high use. Among women, 15 had no service use, 25 had low use, 15 had moderate use, and 20 had high use.

Interviews

Interviews lasted an average of about 60 minutes and were conducted in-person by trained interviewers using a semi-structured guide. The interview guide contained up to 74 open-ended questions addressing health-related practices (alcohol consumption, smoking, diet/weight management, and exercise). The questions asked depended on participants’ responses and on their alcohol consumption (e.g., lifelong abstainers were not asked about current or past drinking practices). Interviewers confirmed current drinking status as part of the interview (lifelong abstainer, current drinker, former drinker). All interviews were audiotaped and transcribed.

Qualitative analyses

We used Atlas.ti (Muhr T, 2004) software to code transcribed versions of all interviews. The study team (including the two study interviewers) developed an initial descriptive coding scheme following review of text from multiple interviews. Once the preliminary coding scheme was developed and codes defined, the codes were applied, discussed, and modified. Once finalized, all code definitions were explicitly defined, and weekly reliability sessions were held in which sections of interviews were coded by all coders and discussed to identify and resolve discrepancies. Three of the authors of the manuscript (CG, MP, SJ), as well as the two study interviewers, coded the 150 transcripts. Intercoder reliabilities were calculated for 9 primary codes in 10% of the transcripts. Overall, 208 passages of text were reviewed and primary coders were judged to have applied codes accurately 92.8% of the time.

Once transcripts were coded, we used Atlas.ti queries to create reports of text from codes that addressed reasons for quitting drinking and reasons for avoiding or seeking health care. Each query was reviewed by two authors to identify descriptive, interpretive, and pattern codes, and to identify themes in the text (Lofland & Lofland, 1995; Luborsky, 1994). Common themes are reported here.

RESULTS

Quantitative analyses

Results of the first set of negative binomial and logistic regressing analyses appear in Table 1, which shows predictors of service use for all respondents using the drinking status measure as the alcohol consumption variable. The improvement in model fit from the intercept-only model (using Negelkerke R²) ranges from a low of 5% for inpatient days to a high of 31% for telephone calls; there was a 26% improvement for primary care visits and 28% improvement in likelihood of or having a mental health visit.

Table 1.

Results of regression analyses predicting service use among survey respondents in the 12 months following survey return. Alcohol measure is drinking status.¹

	Urgent Care Visits (IRR, 95%CI)	Emergency Department Visits (IRR, 95%CI)	Primary Care Visits (IRR, 95%CI)	Telephone Calls (IRR, 95%CI)	Specialty outpatient visits (IRR, 95%CI)	Inpatient days (IRR, 95%CI)	Specialty Mental Health visits (OR, 95% CI)	Specialty Addiction Medicine Visits (OR, 95% CI)	Service Use Outside KP (OR, 95% CI)
Lifelong abstainer	—	—	—	0.70(0.59–0.84)****	0.86(0.76–0.98)*	—	—	—	—
Former drinker	—	1.22(1.04–1.43)**	1.12(1.06–1.19)****	1.25(1.08–1.45)**	—	2.02(1.31–3.11)***	—	1.97(1.17–3.33)**	—
Drinks 30–59 drinks/month	—	—	—	—	—	—	—	—	—
Drinks 60 or more drinks/month	—	—	—	—	0.78(0.64–0.96)*	—	—	2.80(1.38–5.66)**	—
Female Gender	1.22(1.07–1.40)**	0.86(0.75–0.98)*	1.15(1.09–1.21)****	1.45(1.34–1.59)****	—	—	—	0.56(0.35–0.89)**	1.25(1.10–1.42)***
Gender x Lifelong abstainer	—	—	—	1.46(1.19–1.78)****	—	—	—	—	—
Gender x Former drinker	—	—	—	—	—	—	—	—	—
Gender x Drinks 30–59 drinks/month	0.63(0.42–0.94)*	—	—	—	—	—	—	—	—
Gender x Drinks 60 or more drinks/month	—	—	—	—	—	—	—	—	—
White race	—	0.71(0.60–0.83)****	0.92(0.87–0.99)*	—	—	—	—	—	—
Adjusted income	—	—	—	1.02(1.00–1.04)*	—	—	—	—	—
Employed or student	—	—	—	0.90(0.84–0.97)**	—	0.63(0.43–0.92)*	—	—	—
Education level	0.94(0.90–0.99)*	—	0.97(0.95–0.99)**	—	—	—	0.28(0.17–0.38)****	—	1.26(1.19–1.33)****
Dislike visiting doctor	0.92(0.86–0.98)*	0.90(0.82–0.99)*	0.91(0.88–0.94)****	.089(0.85–0.93)****	0.89(0.84–0.85)***			1.53(1.08–2.17)*	0.90(0.82–0.99)*
Concerned Dr. might disapprove of health practices	—	—	—	0.94(0.89–0.98)**	—	—	—	—	—
If sick, one’s own behavior determines getting well	—	—	—	0.97(0.94–0.99)*	—	—	—	—	—
Better to seek professional help than treat oneself	—	—	1.03(1.01–1.05)***	—	—	—	—	—	0.84(0.80–0.89)****
Religious & spiritual beliefs important	1.14(1.06–1.22)****	—	1.04(1.01–1.08)**	1.05(1.00–1.09)*	—	—	—	—	1.27(1.15–1.39)****
Respect & admiration of others important	—	—	—	—	—	—	—	—	0.90(0.81–0.99)*
Self—efficacy in health care settings	—	—	—	—	—	—	—	—	0.97(0.96–0.99)**
Number of days exercised past 7 days	—	—	1.01(1.00–1.02)*	—	—	—	—	—	—
Usual hours sleep	—	—	—	—	0.91(0.85–0.97)**	—	—	—	—
Current smoker	—	—	—	—	—	—	—	3.55(2.22–5.69)****	0.84(0.71–0.99)*
Copes passively	—	—	—	0.91(0.84–0.98)**	—	—	—	—	—
Copes actively	—	—	—	1.15(1.08–1.23)****	—	—	0.35(0.13–0.56)**	—	1.34(1.19–1.52)****
Behavioral health copayment	0.98(0.97–0.99)**	—	0.99(0.98–0.99)****	—	—	—	—	—	1.03(1.02–1.04)****
Outpatient visit copayment	0.99(0.98–0.99)*	0.97(0.96–0.99)****	0.99(0.98–0.99)**	0.99(0.98–0.99)**	0.98(0.97–0.99)****	—	—	—	0.98(0.97—.099)*
Barriers to care (sum)	—	—	—	1.04(1.00–1.07)*	—	—	—	—	1.13(1.07–1.20)****
Have a primary care provider Q14	1.16(1.03–1.30)**	—	1.12(1.06–1.19)****	1.14(1.05–1.23)***	1.13(1.00–1.27)*	—	—	—	1.29(1.12–1.49)***
Diagnosis of alcohol abuse or dependence prior 12 months	—	1.79(1.07–2.99)*	—	1.41(1.11–1.78)**	—	—	—	11.86(6.43— 21.89)****	—
BMI	—	—	1.01(1.00–1.01)****	1.00(1.00–1.01)**	—	—	—	—	0.99(0.98–0.99)**
Negelkerke R2	.10	.10	.26	.31	.17	.05	.28	.27	.07

¹Note:

^*= p <.05,

^**= p < .01,

^***= p <.001,

^****= p <.0001; analyses controlled for age, ambulatory diagnostic groups, self-reported health status (SF-36 general health, mental health, vitality), RxRisk score, depression diagnosis in the year prior to survey return, and months of health plan membership in the 12 months following survey return; variables that were not predictive of any service use measure are as follows: Coping with alcohol or by smoking, frequency of eating breakfast, consuming 5 or more fruits/vegetables per day, believing that good health is a matter of good fortune, indicating that a comfortable life is important, Hispanic ethnicity, being married or living with a partner, indicating that an exciting life is important, frequency of seat belt use.

Results support hypotheses that there are substantial differences among the different types of abstainers, with findings consistent with the “sick quitters” hypothesis. We found that former drinkers use more services, even when controlling for multiple health status measures. They had more emergency department visits, primary care visits, phone calls, inpatient days, and were more likely to have specialty addiction medicine visits compared to light drinkers (those drinking fewer than 30 drinks/month). Conversely, lifelong abstainers used fewer services when compared to light drinkers: They made fewer phone calls to their doctors and fewer specialty outpatient visits. Particularly noteworthy is that drinking at higher levels (60 or more drinks/month) was predictive of only two types of service use: fewer specialty outpatient visits and more specialty addiction medicine visits.

Consistent with previous research, women made more urgent care visits, more primary care visits, phone calls, and were more likely to use services outside the health plan. They made fewer visits to the emergency department, however, and were less likely to have visits in the addiction medicine department. In examining gender by drinking status interactions, we found only one significant interaction, for gender by drinking 30–59 drinks/month for urgent care visits. Here, men did not differ by drinking status, while women who drank 30–59 drinks/month had fewer urgent care visits than those drinking .5–29 drinks/month.

Beyond gender, predisposing factors that were consistently predictive of later service use included higher education level, which predicted fewer urgent care visits, primary care visits and mental health visits, but greater likelihood of using services outside the health plan. More importantly, disliking visiting the doctor predicted use of fewer services across almost all domains—urgent care visits, emergency department visits, primary care visits, phone calls, specialty outpatient visits, and service use outside KP. Interestingly, it predicted more specialty addiction medicine visits, suggesting that this measure may covary with addiction problems in ways that might lead to service avoidance. Consistent with this interpretation, Polen and colleagues (companion paper #1) found that individuals with frequent heavy drinking disliked visiting the doctor. Other reasonably consistent predisposing factors that predicted later service use included indicators that religious and spiritual beliefs were important (more urgent care visits, more primary care visits, more phone calls, and greater likelihood of service us outside KPNW) and active coping (more phone calls, fewer mental health visits, and greater likelihood of using services outside KPNW). Many other attitudinal and health-practice factors were predictive of service use of one type or another, although most were not consistent predictors across service domains.

Enabling factors that were consistent predictors of service use included copayments for outpatient and behavioral health visits. The former predicted use of fewer services for all types of non-specialty outpatient visits, while the latter predicted fewer urgent care visits and primary care visits, but greater likelihood of using services outside the health plan. Another important enabling factor was self-report of having a primary care provider. Those who did have a provider had increased urgent care visits, primary care visits, phone calls, specialty outpatient visits, and greater likelihood of using services outside the health plan. Barriers to care, as might be expected, predicted more telephone calls and greater likelihood of using services outside the health plan. Body Mass Index is difficult to classify in the Andersen model, likely crossing predisposing, enabling, and need definitions. Irrespective of its classification, it predicted increased primary care visits and telephone calls, but reduced likelihood of using outside services.

Coefficients are not presented for most need factors (as we were controlling for those), although we report prior-year diagnosis of alcohol abuse or dependence because one of our foci was alcohol use. Consistent with previous findings, this indicator predicted greater service use, but only for emergency department visits, telephone calls, and likelihood of specialty addiction medicine visits.

Results of regression analyses evaluating the effects of hazardous or harmful drinking (those positive on the AUDIT) among current drinkers appear in Table 2. In general, the patterns of these results are similar to those reported when considering most predisposing and enabling factors, but they differ considerably when the drinking measures are considered. In these analyses, a score of 5 or more on the AUDIT predicted fewer urgent care visits, primary care visits, telephone calls, specialty outpatient visits, and greater likelihood of having specialty addiction medicine visits. This is generally consistent with the hypothesis that heavy drinkers avoid seeking medical care, yet these results, when compared to those from analyses of drinking status, suggest that it is not heavier drinking, per se, that results in service avoidance, but rather that it is at-risk drinking that results in care avoidance.

Table 2.

Results of regression analyses predicting service use among survey respondents in the 12 months following survey return. Alcohol measure is scoring positive on the AUDIT.¹

	Urgent Care Visits (IRR, 95%CI)	Emergency Department Visits (IRR, 95%CI)	Primary Care Visits (IRR, 95%CI)	Telephone Calls (IRR, 95%CI)	Specialty outpatient visits (IRR, 95%CI)	Inpatient days (IRR, 95%CI)	Specialty Mental Health visits (OR, 95% CI)	Specialty Addiction Medicine Visits (OR, 95% CI)	Service Use Outside KP (OR, 95% CI)
AUDIT Positive	0.79(0.69–0.91)***	—	0.94(0.89–0.99)*	0.92(0.85–0.99)*	0.88(0.78–0.99)*	—	—	2.00(1.11–3.59)*	—
Female Gender	1.14(1.01–1.29)*	0.75(0.63–0.89)***	1.12(1.06–1.19)****	1.40(1.29–1.59)****	—	—	—	0.50(0.30–0.85)*	1.30(1.12–1.51)***
Gender x AUDIT Positive	—	—	—	—	—
White race	—	0.73(0.59–0.89)**	0.91(0.84–0.98)*	—	—
Hispanic ethnicity							0.27(0.08–0.91)*	—	—
Education level	0.93(0.88–0.99)*	—	0.96(0.93–0.98)***	—	—	—	—	—	1.27(1.18–1.36)****
Dislike visiting doctor	0.92(0.84–0.99)*	0.84(0.75–0.95)**	0.88(0.85–0.92)****	0.87(0.83–0.93)****	0.89(0.81–0.97)**
Better to seek professional help than treat oneself							—	—	0.84(0.79–0.89)****
Concerned Dr. might disapprove of health practices	—	—	—	0.93(0.88–0.98)**	—	0.65(0.50–0.86)**
If sick, one’s own behavior determines getting well	—	—	0.97(0.95–0.99)**	—	—
Exciting life important							—	—	1.25(1.12–1.40)****
Religious & spiritual	1.19(1.10–1.30)****	—	1.04(1.00–1.07)*	1.07(1.01–1.02)**	—
beliefs important
Consumes 5 or more fruits/vegetables per day							—	—	1.26(1.07–1.48)**
How often use seat belts	—	—	—	0.92(0.85–0.99)*
Number of days exercised past 7 days	—	—	—	—	—	—	—	—	1.04(1.01–1.08)*
Current smoker	—	—	—	—	—	—	—	2.59(1.36–4.92)**
Copes actively							1.61(1.23–2.09)****		1.32(1.14–1.53)****
Copes passively	—	—	—	—	—
Outpatient visit copayment	0.98(0.97–0.99)**	0.97(0.96–0.99)***	0.99(0.98–0.99)**	0.99(0.98–0.99)*	0.97(0.96–0.99)****
Behavioral health copayment							—	—	1.03(1.01–1.05)****
Barriers to care (sum)	—	—	—	1.04(1.00–1.08)*	—	—	—	—	1.13(1.05–1.22)***
Have a primary care provider	1.15(1.00–1.32)*	—	1.15(1.07–1.23)****	1.20(1.10–1.32)****	—	—	—	—	1.23(1.04–1.46)**
Diagnosis of alcohol abuse or dependence prior 12months	—	—	—	1.58(1.19–2.10)**	—	—	—	6.26(2.31— 16.95)****
BMI	—	—	1.01(1.00–1.01)**	—	—	—	—	—	0.99(0.98–1.00)*
Negelkerke R2	.09	.09	.24	.27	.14	.04	.25	.21	.08

¹Note:

^*= p <.05,

^**= p < .01,

^***= p <.001,

^****= p <.0001; analyses controlled for age, ambulatory diagnostic groups, self-reported health status (SF-36 general health, mental health, vitality), RxRisk score, depression diagnosis in the year prior to survey return, and months of health plan membership in the 12 months following survey return; variables that were not predictive of any service use measure are as follows: Coping with alcohol or by smoking, usual hours of sleep, frequency of eating breakfast, Self-efficacy in health care settings, believing that good health is a matter of good fortune, indicating that a comfortable life is important, indicating that the respect & admiration of others is important, adjusted income, being married or living with a partner, being employed or a student.

One interesting difference from the first set of analyses was that active coping, among drinkers, predicted increased likelihood of having mental health visits, while it predicted fewer such visits when all respondents were included in analyses. Finally, it is worth noting that there were no significant gender by AUDIT-positive interactions.

Qualitative analyses

Our qualitative data support and extend our quantitative findings, providing possible reasons for service use not explained in statistical models. First, we examine the extent to which health affects decisions to stop drinking (addressing the “sick-quitters” hypothesis). Second, we assess the ways in which health-related practices affect willingness to seek care (addressing the hypothesis that individuals with poor health practices, particularly heavy alcohol consumption, avoid care). Lastly, we examined other reasons for avoiding health care as well as reasons for seeking such care.

Health and quitting drinking: Assessing the “sick-quitters” hypothesis

Consistent with our quantitative finding that former drinkers used more health care, about half of the former drinkers who gave reasons for stopping drinking said that health played a role in their decision to quit. In some cases participants gave specific health examples that helped motivate them to quit, while in other cases participants mentioned vague health issues related to alcohol (e.g., “I didn’t like the way I felt when I drank”). Only 3 of the people we interviewed specifically said that health was not a factor. The following examples provide indicators of the specific types of health problems that motivated abstinence among former drinkers.

It was presented to me 10 years ago in a way that I was abusing myself. I was drinking and stuff, and I got a hangover and it didn't go away. After a week I went in and they said that I was spilling sugar so bad I should have been in a coma. The doctor told me that I will quit drinking, smoking, change your diet and do your medication right or you will die in a couple of years. Presented that way, I did it.

Interviewer: Did health-related issues have anything to do with your decision to stop drinking?

Participant: Yeah. I knew my liver enzymes were high. I know what alcohol can do to me, just physically; never mind the mental side of it. Cancer. I had already had chronic gastric problems, which I knew were related to the amount of alcohol I drank.

These explanations suggest that individuals stop drinking in response to health problems, and that at least some of these health problems are directly related to alcohol consumption, as hypothesized.

Avoiding and seeking care: Underlying reasons

Effects of lifestyle and health practices

We found two primary cross-cutting themes related to reasons underlying care avoidance due to health-related practices. The first was feeling embarrassment and shame. The following quotes illustrate how embarrassment and shame across various health practices affected care seeking:

…I didn’t seek medical attention because I was embarrassed…well it was related to female health issues and my lifestyle at that point in time wasn’t exactly on the up and up, and I felt I should be living a different life than I was, but I didn’t really want anybody to tell me that. I knew that, but…I didn’t want the commentary. I didn’t want the lecture and I didn’t want the like, “Oh, my god” They think I’m a slut.”…so I didn’t seek attention when I probably needed to.

Right before I decided to lose weight I was really feeling pretty ashamed of what I was looking like…I just felt like I was so ashamed of myself that I didn’t want to hear someone else tell me that I needed to do it. I already knew that I needed to do it [lose weight].

Interviewer: Did you have any concern about what a doctor might say to you about your drinking, when you were drinking?

Participant: …Towards the end, when I was getting pretty disgusted with it myself, yes, I was embarrassed by it. Before that, no.

I don’t want to take off my clothes in front of any of the doctors. I don’t think the gowns that they give you to wear are very big. They are tiny and they are short…for a fat person.

I was ashamed about it. I didn’t want to be known as one of those people that stunk and smoked cigarettes and people looking at me differently. I just didn’t want to be looked at that way.

The second theme related to care avoidance because of health practices was recognizing that a health behavior was causing a health problem, knowing that the doctor would suggest behavior change and, sometimes, not wanting to hear about the behavior. As one woman said, “I’m pretty aware of everything bad that I do. I don’t really need to go to the doctor to have that brought up. I know it’s bad.” The following quotations provide examples across various health practices:

...just because of my weight I just assume that what they’re going to tell you is to lose weight…

I quit smoking in September and right before that I was really ill. It was in my lungs and I felt like it was my smoking and I was a little apprehensive about seeing the doctor, because I just knew that she was going to say that I just need to quit smoking. I was just at the point where I didn’t want to hear that. I knew that.

Participant:…I guess they knew that I was drinking too much, so when somebody points it out to you, it makes you feel uncomfortable.

Interviewer: Did that have any bearing on your willingness to go back into seeing the doctor in the future?

Participant: No, but it crossed my mind.

A related sub-theme was lying to clinicians about health-related practices to avoid embarrassment or to avert discussions or lectures about the behavior:

I think it has to do with not wanting to share things about yourself that you're not proud about. If I went in because I went in for a sinus infection and I didn't think my drinking had anything to do with that, but if the doctor asked about it I might not have told him to the full extent just because I wouldn't want to put myself in a bad light.

Interviewer: When you were chewing [tobacco], were you ever reluctant to go to the doctor for any reason?

Participant: No, I would lie about it...

Interviewer: What was it that prompted you to lie?

Participant: That they would lecture me. I didn’t want to deal with it.

Disliking the care process

Consistent with our quantitative findings, people reported avoiding care because they disliked seeing the doctor. They reported feeling uncomfortable in the clinical setting, disliking particular procedures, experiencing dread or fear, or just disliked the process necessary to seek care—whether it was related to the rushed nature of appointments, the waiting or clinic room facilities, or the time it took to make an appointment. Particularly uncomfortable procedures—primarily prostate exams, colonoscopies, sigmoidoscopies, or mammograms—also produced avoidance, as did dental work.

Interviewer: Have there been times when you have put it off or delayed going or decided to not see a doctor at all when you think you probably should have?

Participant: Oh yeah…I don’t like doctors…Even my primary care physician. He and I have a good understanding. I don’t like doctors and he understands that and he works with that. He knows that when I come in that I am not the easiest patient he is going to have that day—I can promise you that—but he deals with me at my level…we get along fine.

Interviewer: Is it related to what you just mentioned about the discomfort in the office and the doctor being in a hurry?

Participant: Yeah, pretty much. I think so. I would just rather not go. Yeah. The little rooms are so clinical and it feels pretty rushed.

Participant: I’m going through a thing right now where they want to do—I forgot the special medical terminology…I went in and it was too painful—and it’s for my bottom. They got like three quarters of the way up and it was just too painful. Huh uh!...

Interviewer: Sigmoidoscopy or a colonoscopy?

Participant: Yes, yes. Yeah. Nah! Skip that. I’m sorry, I did that much and you ain’t getting no more out of me on that. Huh uh!

Interviewer: So what is it that keeps you from getting there?

Participant: I don’t know, I really don’t. I dread going. I really do. You don’t know what to expect is what I think it is. I wait until the last possible minute to do it. Normally it’s because somebody else on the other side of the door is pushing me to go.

…if there is a reason why people don't go to the doctor, is it for a specific area. Like mine, it's, "I don't want a mammogram and I don't want a Pap Smear." He can do anything else to me—well almost anything else. [Both laughing.]

Interviewer: What kept you from going [to the doctor]?

Participant: Scared. Fear…of what might be wrong with my body…I do go, but I stay away as long as I can.

Other reasons provided for not seeking care were forgetting or being too busy to go, not wanting to admit to having a problem like depression, not recognizing that they actually had a problem, and not wanting to waste the doctors’ time or feel that they are being perceived as a hypochondriac by physicians and clinic staff. These latter individuals don’t want to “cry wolf.”

…It seems like I come to the doctor a lot and I just didn't want people thinking, oh, here she is again. What is it this time? So I've held back until the issue got worse and then I came in.

Seeking care when illness is significant or interferes with activities

Many of the people we interviewed did not avoid care, and of these, a significant number said they would want to know if anything was wrong with their health. Many also tended to “watch and wait.” This approach was characterized by waiting to see if a known health problem gets better on its own or with home treatment, before seeking formal medical care. Problems perceived as more serious were seen differently from those that were less serious. Overall, people often did not seek care unless something appeared to be seriously wrong, or began interfering with functioning, such as with work:

…if you're talking about your heart and diabetes, and that, anybody with, I think, an iota of smarts would go. But if we're talking about a mole or your finger aches, something that you don't look exactly hunky dory, but you think, well, it might get better…I might not be so quick to jump on it if it's, again, just something that I think either I can fix, or time will fix it.

Faith that the body will heal, that the body is healthy, or believing that the doctor cannot do anything about the problem

Some people have a strong faith in their body’s ability to heal itself when a medical issue arises, or feel that they are healthy and that serious problems are not likely to occur. Many also recognize that there are problems that health care providers cannot address.

Interviewer: Okay. Has there ever been a time when you delayed going to the doctor or didn't go to the doctor at all even though you probably should have?

Participant: Sure, because I haven't gone in for eight years for another physical. I've said to myself that I'm not going to ask him for a stress EKG because he is just going to laugh. Laugh as in they are just going to go, what do you need that for? …You ride your bike all the time. You can do anything that you want. You do high exercise and run, and whatever you want to do, and you don't have any problem. So, yeah, I am currently delaying seeing the doctor because I don't have an acute issue.

Interviewer: Have you ever not gone to the doctor, or avoided going because there was any kind of fear or anxiety that you might learn that you had some serious health condition?

Participant: No. It's never really crossed my mind, I guess. It's just never occurred to me that's there's something wrong; that it could be REALLY something wrong.

Delaying rather than avoiding care

When people did delay or avoid care, it was often for a short period rather than indefinitely. Others reported that they had delayed or avoided care earlier in their lives but not as they grew older, most often because they either recognize that earlier intervention is better for them in the long run or because they figure it is more likely something will be wrong as their age increases.

I'm human, so I'm quite sure there has been some time when I was younger [when I avoided care] but not recently…I'm 40 years of age and I'm in that high risk [racial minority] group…I'm quite sure that when I was in my thirties I probably didn't go. I probably thought it will be all right, or that I could wait until the next week and then finally it got to the point where I had to go.

Encouragement of others

Consistent with our finding that being married or in a committed partnership predicts increased use of preventive services, particularly for men, some participants reported that the encouragement of others was the cause of their service use:

When I was having those heart problems I didn’t go in. I didn’t want to. I’d put things off. I’m a procrastinator. I almost died one time because I had a bee sting and I had a reaction. …If my wife hadn’t nagged me, I’d have died.

Effects of fear on care seeking

Fear affected care seeking but the effect differed in important ways across interviewees: some avoided care entirely, some merely delayed, others actually sought out care in response to their concerns. Some participants mentioned fears surrounding their perceived increased risk of getting illnesses due to a family history. When people mentioned fears related to a family history of an illness, about as many reported they were more likely to go to the doctor or be extra vigilant about their own health as those who said they avoid the doctor because of such concerns. Many people also overcame their fears and saw the doctor, although the process of overcoming fears could delay care seeking. The following example illustrates the ways that participants discussed their fears:

Interviewer: Have you ever avoided going to the doctor or been reluctant to go …because you were afraid you might have a serious illness?

Participant: There has never been a time when I thought I had a specific serious illness, but I think that's part of getting a checkup in general. It's that I'm afraid. This time I had blood work done and everything, which I hadn't for a really long time. I think that might be part of why I don't go. I think it's this fear that I might have leukemia or something like that. But also I think it's important that you know those things. That is also one of the reasons that I do go.

Interviewer: So in the end, the need to know overrides the fear?

Participant: Yeah, I think so. I think it's important to know.

Barriers to care seeking

Finally, people mentioned various barriers to seeking care. These included lack of insurance and wait times for appointments. Most often mentioned was lack of money. Barriers related to dependent care were also discussed.

DISCUSSION

Our quantitative results provide strong evidence that health-related practices and attitudes predict subsequent use of health care. Of these, the most consistent predictors of later care-seeking are having quit drinking alcohol, at-risk alcohol consumption, cigarette smoking, greater BMI, disliking visiting the doctor, and indicating that religious or spiritual beliefs are important.

We found that under most circumstances, it is not heavier alcohol consumption per se that affects individuals’ willingness to seek health care services, rather, it is hazardous drinking and drinking-related problems that predict service avoidance. These analyses suggest that it is when individuals recognize that they are identifiable as being at-risk for drinking-related problems that they begin to avoid seeking care. In other words, it appears that individuals who know they are drinking in ways that place them at risk are less likely visit their health care providers. Consistent with this finding are those in Polen and colleagues (companion paper #1), indicating that individuals with frequent heavy drinking dislike visiting the doctor more than others do.

These results are also interesting in comparison to our findings on predictors of preventive services use (Green et al., companion paper #2). It appears, for example, that smokers avoid preventive care but not routine care, while at-risk drinkers avoid routine care (e.g., illness-related care) but not preventive services. This suggests that the relationships between health-related practices and different types of care need to be understood differently, particularly with respect to stigmatized health-related behaviors.

Our qualitative analyses provide information critical to understanding why these factors affect service use, and also suggest opportunities for changing the delivery of care in ways that might reduce care delay and avoidance. We found that embarrassment and shame are strong motivators of avoidance for individuals who do not follow recommendations for health practices or weight maintenance, and addressing these concerns may be fruitful. Similarly, simple techniques for improving clinicians’ communications about difficult topics might benefit patients who do not want to be lectured or hear more about their problematic health practices. Other areas for improvement include addressing processes of care that make people uncomfortable—working to make the care environment more hospitable (physically and in terms of interactions with care providers), and making difficult or painful medical procedures more comfortable.

Finally, although we included a host of health, functional status, attitudinal and behavioral predictors in our models, they remained similar in fit to other models of service use.

Limitations

The survey response rate was lower than desired, suggesting the possibility of bias in our sample. We were able to compare respondents to non-respondents, however, providing information describing differences that might affect our results. A strength of the study, however, is that our service use data are prospective and derived from health plan data rather than self-report.

CONCLUSIONS

Our findings suggest that the factors we typically expect to affect service use remain inadequate in explaining the majority of the variance in such use. In this context, our qualitative analyses identified various reasons for seeking or avoiding care that are not typically measured in studies of health care utilization. These include embarrassment and shame, fear, faith that the body will heal, lack of expectation that anything significant will go wrong, disliking the care process, the need to understand a perceived health risk or problem, and seeking care only when experiencing a significant illness or when functioning is affected. Future efforts might measure these factors to help increase our ability to understand and explain the factors that lead to and prevent use of health care.