Variation in dental treatment – Why?

	Oral Health Care Delivery
	November 5, 2003

Variation in clinical decisions

	Widely acknowledged that there are dif-ferences in:
		how dentistry is practiced among regions in the country
		Differences among practitioners in the same area
	Has been infrequently studied

Variation in clinical decisions

	These differences commonly accepted as reflections of …
	1) The “art of dentistry”
		The natural variation in the best clinical judgment of dentists concerning individual patients
	2) Uncertainty or disagreement concerning the most effective approach to treatment

Variation in clinical decisions

	Advocates of the latter (#2) argue that:
	If two dentists consistently provide a different set of preventive and treatment procedures for patients with similar conditions
	Then one dentist must be providing less effective care than the other
	Unless the care leads to equivalent results for patients when compared across a wide range of possible outcomes

Variation in clinical decisions

	Variation in medical practice has been studied for two decades
	Led to investigations of
		under- and over-utilization of medical care procedures
		And of the evidence for the effectiveness of some treatments
	This course of inquiry has begun to mature with the develop of methods to designed to improve clinical decision making, such as practice guidelines

Variation in clinical decisions

	Few studies have reported quantitative measures of the extent of agreement among dentists in their caries diagnoses in vitro

Treatment recommendations for individual teeth made by 15 dentists for the same patient

	An attempt to quantify the extent of agreement among dentists
		for decisions to recommend treatment
		and to recommend treatment for reasons related to caries

Treatment recommendations for individual teeth made by 15 dentists for the same patient

	Table 1 – shows the recs for treatment for one patient by 15 different dentists
	Each of the 13 teeth receiving one or more recs for tx appears as a row in the table
	For each tooth, the number of dentists recommending tx overall is shown
	As well as the number recommending treatment for each of four mutually exclusive reasons:
		Primary caries        Secondary caries
		Faulty restoration   Other

Treatment recommendations for individual teeth made by 15 dentists for the same patient

	For only one of the 13 teeth receiving at least one rec for treatment was there “perfect agreement” among the 15 examining dentists
	Of course, perfect agreement not to recommend treatment was achieved for 15 additional teeth not shown in the table
	For all but one of the teeth with multiple recs for tx, there is disagreement about whether caries is the principal reason for treatment

2.  Intraclass correlation coefficient (rho) for agreement among dentists’ recs for restorative tx due to caries, for teeth grouped by restoration and caries status

	Second measure of agreement was a modified intra-class correlation
	Where total variance attributable to differences among examiners for individual teeth is divided by the total variance across examiners and teeth
	Using these two measures, agreement on treatment recommendations due to caries was calculated for 1,187 teeth, and for four subsets of teeth grouped by caries and restoration status, as determined by a visual-tactile epidemiological examination

2.  Intraclass correlation coefficient (rho) for agreement among dentists’ recs for restorative tx due to caries, for teeth grouped by restoration and caries status

	The rho value of 0.53 is in the “fair” half of the “fair to good” range
	When outcomes examined for teeth grouped by the presence or absence of a restoration and of clinically evident caries, clear that the presence of a restoration is associated with a deterioration in the extent of agreement
	both when clinically evident caries is present
	and when it is absent.
	Agreement only enters the good range for teeth with clinically evident caries and no restoration

2.  Intraclass correlation coefficient (rho) for agreement among dentists’ recs for restorative tx due to caries, for teeth grouped by restoration and caries status

	Just as was suggested for the 13 teeth in Table 1
	some of the lack of agreement may be due to differences in attribution of reasons to recommend treatment
	Rather than difference in decisions to recommend tx
	When parallel analyses were performed to determine extent of agreement for treatment recommended for any reason
	the rho values were approximately 0.10 higher for each tooth status category except unrestored carious teeth
	Differences in extents of agreement
		as well as the deterioration caused by the presence of a restoration
		suggest that variation among practitioners is elevated in assessments of the caries status of restoration margins

Variation in decision-making – Phases of decision-making

	The clinical decision-making process in dentistry can be divided into three separate phases
	Differences in dentists’ decision making that results in variation can arise within any of these three phases
	1.  diagnosis, or more usually the detection phase, where a disease is identified
	2.  the decision about intervention, given that a disease or condition is identified
		Usually a yes/no decision, although dentists routinely indicate uncertainty by noting in the patient record that  a condition by “watched”
	3.  selection of the treatment from among alternatives, if a decision to intervene is made

Variation in decision-making – Diagnostic phase

	Likely that differences shown in tables
	Reflect 4 types of variation in application of diagnostic criteria:
		Disagreement on:
			the presence/absence of a condition
			terminology for a given clinical condition
			whether an identified condition is carious
			whether an identified carious lesion or other condition requires treatment

Variation in decision-making – Diagnostic phase

	Clearly dentists do exhibit substantial differences in the diagnostic or detection phase of clinical decision making
	Likely that two factors account for such difference:
		Skill and diligence in the examination
		Definition and criteria employed for the identification of a disease

Variation in decision-making – intervention phase

	The decision about intervention
	Given that a disease or condition is identified:
		Usually a yes/no decision
		although dentists routinely indicate uncertainty
			by noting in the patient record that  a condition is being “watched”

Variation in decision-making – treatment choice phase

	Selection of treatment from among alternatives
		has had least attention devoted to it
	Lack of attention may be a result of:
		limited number of alternatives available for some conditions
		lack of a perceived need to evaluate alternatives
				in face of strong beliefs about their relative effectiveness

Variation in decision-making – treatment choice phase

	Strongly held beliefs
		an extremely important determinant of treatment selection, often in the absence of any supporting evidence
		Belief  that crowns are superior to amalgams
			But no definitive survival analyses
			Or examinations of cost effectiveness are available to support this belief
			And dentists’ estimates of the longevity of amalgams and crowns vary widely
		Placing amalgams rather than sealants in fissures
		Also associated with strongly held beliefs about relative effectiveness
			This time in the face of evidence for the effectiveness of the less invasive procedures

Investigating reasons for variation

	Investigation of factors associated with variation among dentists is infrequent
	Variation introduced by differences in patients

Slide 20

Investigating reasons for variation

	Controlling patient factors:
		through selection of practices with similar patients,
		presenting the same patients to multiple practitioners
		presenting “teeth” to multiple practitioners
	necessary if nonpatient factors are to be identified with any certainty

Investigating reasons for variation

	Type and/or extent of variation among dentists may be different
		depending on characteristics of patients examined, such as:
	Previous treatment history
	Age
	Insurance status

Investigating reasons for variation

	When these patient factors are controlled -
	Variation in clinical decisions
		must be ascribed to differences in clinical decision making among dentists

Delivery System within which tx provided –Source of Variation

	Incentives emphasized by the system
		Currently reflect differences among patients more than differences among dentists
		Delivery system may become a principal dentist characteristic in the future
	financial incentives & constraints imposed by financing mechanisms
		May contribute to differentiate among groups of dentists operating under assorted reimbursement arrangements

The Clinician as Service Deliverer -
Clinician characteristics – variation source

	The “delivery system” as the provider located within his/her practice setting
		Age
		Gender
		Experience
		Type of training
	Personal factors, help to explain why variation among clinicians exists even when clinical situation is similar

The Clinician as Service Deliverer –
Style of practice – variation source

	Evidence for styles of practice
	Characteristics of the patients whom practitioners want to have in practices
		On basis of age, race, sex, or social status
	Preference to care for certain types of clinical problems
		Unusual, challenging  vs. easy to manage
		Convenience
		Availability of laboratories, etc.
		Encourage or discourage follow-up visits, referrals

The Clinician as Service Deliverer
The practice setting – variation source

	Influence of fellow professionals
		can be particularly strong
	can produce group practice style
		peer pressure
		professional leadership

The Clinician as Service Deliverer
 Role of Professional Leadership – variation source

	Influence of professional leadership –
		set standards
		peer pressure
		professional leadership
	Certain individuals seen as influential
		Lit on health services research shows that influence of professional leadership has been prominent in physicians’ acceptance of innovations

The Clinician as Service Deliverer
Income earner – variation source

	Some evidence that clinicians’ desire for income has influence on practice patterns
	Derives from studies of the relationship between the supply of clinicians and the cost and volume of practice
	The literature on clinician-induced demand implies that clinicians can create demand to generate income
	There is evidence that clinicians do adjust the service they provide in response to economic incentives
	Even if do not induce new demand, may substitute one of their services for another

Differences in how dentists Interact with Patients  - Variation source

	Dentists may range at extremes of interaction styles:
		from authoritarian prescriber
		to patient counselor
	thus varying extent to which patient preferences influence decisions

The Clinician as Patient’s Agent
Patient’s economic agent – variation source

	Price patient must pay for services seems to affect physicians’ patterns of prescribing services
	Physicians respond to differences in price of medical care by varying their utilization of services
	Number of tests ordered seems to be influenced by cost

The Clinician as Patient’s Agent
The Patient’s Clinical Agent – variation source

	Serve as patients’ agent in ensuring the quality of care and the provision of services necessary
	Most important variable in choice of drugs is the perception of their effectiveness and risk
	Although price influences clinicians’ use of dx test, clinical considerations substantially more important

The Clinician as Patient’s Agent
The Patient’s Clinical Agent – variation source
Dealing with uncertainty

	Many have described the importance of clinical uncertainty in clinicians’ practice patterns
	Because clinicians’ beliefs about the effectiveness and the risk of clinical services is of central importance
		The certainty with which they hold these beliefs is critical in determining their decision-making

The Clinician as Patient’s Agent
The Patient’s Clinical Agent – variation source
Dealing with uncertainty

	Eddy has emphasized the ambiguity inherent in defining the difference between
		Normal and abnormal
		Characterizing disease entities
		Collecting accurate data
		Evaluating diagnostic tests
		Measuring outcomes

The Clinician as Patient’s Agent
The Patient’s Clinical Agent – variation source
Dealing with uncertainty

	Uncertainty may exist because knowledge simply not available
	Or because do not have access to the available information about the usefulness and risk of a service
	Or because are unable to make use of the available information appropriately

The Clinician as Patient’s Agent
The Patient’s Clinical Agent – variation source
Dealing with uncertainty

	Because uncertainty seems inevitable – must be considered as an element in clinical decision-making
	John Wennberg has identified three sources of uncertainty:
		Difficulties in classifying a particular patient, so that probabilities of disease, extent of disease, prognosis and treatment outcomes cannot be reasonably ascertained

The Clinician as Patient’s Agent
The Patient’s Clinical Agent – variation source
Dealing with uncertainty

	Sources of uncertainty (cont.)
	2.  Information commonly does not exist on the probabilities of treatment outcomes under controlled circumstances
	3.  Uncertainty exists because the values of the patient may not correspond to the patient’s values

The Clinician as Patient’s Agent
The Patient’s Clinical Agent – variation source
Uncertainty

	Sources of uncertainty (cont.)
	4.  Clinical care is simply a risky business –fact that outcomes expressed as probabilities
	means there is uncertainty about how an individual patient will turn out
	Uncertainty in clinical practice includes more than lack of knowledge
	Even if group statistics are known, the outcome will continue to be uncertain for each patient
	as long as clinical care is a probabilistic process

The Clinician as Patient’s Agent
The Patient’s Clinical Agent – variation source
Uncertainty

	Way the clinician deals with uncertainty can influence the amount of variation
		Establish routines
			(local clinical rules of thumb often guide practice in areas of high uncertainty)
			As a result  are wide interregional differences in practice patterns
		Uncertainty enters through the patient
		Erring on the side of conservatism
			Defensive medicine?

Implications of variation in treatment

	Unexplained variation in practice style:
		suggests that some care may be inappropriate.
		when explicit criteria applied:
			instances of both overutilization & underutilization found

Implications of variation in treatment

		Even when differences in patients’ preferences
		case mix
		severity of disease considered
	there are potential improvements in the cost and quality of care that might result from changes in the practice habits of clinicians

How to change behavior?

	Task of changing behavior is a daunting one
		Clues from behavior modification
		Ideas from management theory
		Lessons from adult learning theory

Do Education and Feedback change clinician’s decisions?

	Will education change practice?
	The role of professional leadership in education
	Education and perceived need
	Does the change persist?
	Is the medium more important than the message?