Politics is certain to determine the status of routine HIV screening; you (the physician) are more likely to be required to submit to screening than is your surgical patient. There are supposedly about 300,000 people in the US who are HIV positive but don't know it (Arch. Int. Med. 162: 887, 2002). Since the first ten years of the epidemic, the rhetoric has shifted from "protecting people's right not to be tested" to "getting people in for voluntary testing" and bewailing the fact that most primary care physicians don't screen people at risk (Am. J. Pub. Health 92: 1784, 2002). Probably you still can't test without the person's consent (even in prisoners: Pub. Health Rev. 116: 520, 2001), so now hospitals serving underclass communities have teams to get inpatients to accept testing (Arch. Int. Med. 162: 887, 2002). Especially, the availability of medications to keep Baby from catching HIV from Mom has made screening important, and even most (not all) ethicists consider that society may require Mom to accept treatment. OraSure has revolutionized HIV testing among the underclass (Am. J. Pub. Health 94:29, 2004), and PCR for the virus testing is (or should be) routine so as not to miss the acute infections (JAMA 288: 216, 2002).

Curiously, the "ethicists" are only now starting to notice that we've been testing for drugs of abuse for decades without informed consent (Am. J. Med. 115: 54, 2003); I predict this will go nowhere because people's lives in an emergency setting depend on our being able to get this information without any interference from persons of delicate conscience. (The person screened is still protected by the rules of physician-patient confidentiality and the requirement that if a test is to have a legal impact, it must be confirmed by a separate test using a different chemical principle.)

What was supposed to be the modern era's "last word" on screening was the 1989 United States Preventive Services Task force (USPSTF; available from William & Wilkins). It's already dated. For an older overview of screening, see JAMA 254: 1480 and 1499, 1985. Future surgeons see JAMA 253: 3576 and 3589, 1985; Surgery 109: 236, 1991; Arch. Path. Lab. Med. 115: 1088, 1991 ("routine bleeding times"); Am. J. Surg. 188: 671, 2004 (routine labs for injured kids are worthless); Ob. Gyn. 78: 837, 1991 (PT before elective gynecologic surgery is a waste), Pediatrics 89: 691, 1992 (routine screening of kids for coagulation problems is plus-minus helpful); Milit. Med. 154: 140, 1989, Ann. Int. Med. 114: 432, 1991, Mayo Clin. Proc. 66: 155, 1991, Arch. Surg. 127: 801, 1992 ; Surgery 115: 56, 1994; Med. Clin. N.A. 87: 7, 2003 (the last six all agreed that routine screening of healthy people before elective surgery is a waste; these articles also cover the clinicians' butts should their failure to order labwork miss a cholesterol or calcium problem or something). Application of USPSTF guidelines to the elderly (with "cost-effectiveness" data): J. Fam. Pract. 34: 320, 1992.

Law mandates the screening of newborns for such things-requiring-early-recognition as hemoglobinopathies, phenylketonuria, galactosemia, and congenital hypothyroidism (Arch. Dis. Child. 67: 1073, 1992). Many states also mandate screening for congenital adrenal hyperplasia, biotinidase deficiency (limb paralysis and blindness, controllable by giving biotin), G6PD deficiency, Duchenne's, and/or maple-syrup urine disease (progressive brain damage, preventable by limiting intake of branched-chain amino acids). Today's tests are so sensitive (which is good) that there are about 50 false-positives for every case found (Arch. Ped. Adol. Med. 154: 714, 2000). These probably get done automatically. Neonatal screening and the problems caused by false-positives: Clin. Chim. Acta. 315: 99, 2002; Arch. Ped. Adolesc. Med. 154: 714, 2000. You will also screen young kids for lead poisoning (Am. J. Pub. Health 93: 1253, 2003); old lead paint in houses remains a major health threat about which the politicians are doing nothing (contrast the "asbestos in the schools" hoopla of the late 1980's).

Likewise, protocols (or at least recommendations) exist for lab testing as part of pre-natal care -- everything from blood pressure to glucose to illegal drugs (Am. J. Ob-Gyn. 166: 588, 1992) to alpha-fetoprotein to track down neural tube detects (Am. J. Clin. Path. 97: 541, 1992).

Future clinicians take note: Here's a reasonably current list of things that are probably worth doing routinely for asymptomatic clean-living grown-ups without family histories of anything bad or possible exposures to slow-acting poisons. Ask the local USPSTF guru:

• urging people to turn their hot water temperature down to 120 F;


• urging people not to smoke in bed;
• urging people to install smoke alarms;
• trying to find out who's a problem drinker (good luck!);


• ask the women about domestic violence (Mayo Clin. Proc. 73: 271, 1998)


• urging people to get regular dental care;





• urging people to exercise;
• urging people to wear seat belts;


• urging cyclists to wear helmets (Am. J. Pub. Health 84: 653, 1994)


• discussing general nutrition;


• trying to find out just how many drugs the person is taking;


• recommending cessation of smoking;


• urging monthly breast self-exam for women;


• urging advance directives, durable power of attorney;


• urging an aspirin a day (to prevent strokes and heart attacks; maybe, wait'll you're in practice for a better assessment of benefit/risk ratio);


• consider urging post-menopausal women with hysterectomies (and perhaps without) to take supplemental estrogen;


• offering 'flu shots yearly to everyone 65 and over;


• offering pneumococcal vaccine once to everyone age 65 and over;


• tetanus toxoid every ten years (or diphtheria-tetanus vaccine)


• office health-maintenance visit maybe yearly (maybe more in the old-old); this includes a basic physical exam (remember to look around in the mouth, especially if it's an old smoker: Cancer 72(S3): 1061, 1993;

• check every guy for hernias and testicular masses;


• do a rectal (and check prostate, if a man; Br. Med. J. 300: 1041, 1990);


• checking blood pressure;


• annual clinical breast exam and mammogram to age 75 (USPSTF rec); some people say to start yearly mammograms at age 40 (Ann. Int. Med. 122: 534, 1995; Cancer 75: 2412, 1995; NCI now says no, ACS and the American College of Radiology say yes CA 44: 248, 1994; South. Med. J. 91: 510, 1998); nobody questions their value after age 50 (Cancer 71(S1): 231, 1994)


• pap smears in women (start with first intercourse; do every 1-3 years, stop at age 65 if the last two were normal; Am. Fam. Phys. 45: 143, 1992; Postgrad. Med. 101: 207, 1997);



(New directions in the pap smear industry, including realistic assessment of the high false-negative rate for human slide-readers and introduction of image analysis technology: Cancer 71(S4): 1406, 1993; probe for the four worst HPV strains on routinely-collected material J. Clin. Path. 45: 385, 1992; the tort system has made pap smears unavailable for the women who need them most, while we look to computers to help get rid of false negatives: Arch. Path. Lab. Med. 121: 287, 1997; Arch. Path. Lab. Med. 121: 282, 1997.);

• stool occult blood yearly (old recommendation: every two years ages 40-50, then yearly), maybe a flex-sig every three years (longer than that is probably too long: JAMA 290: 41, 2003);



(Colon cancer screening: Cancer 70(5S): 1246, 1992; Am. J. Gastroent. 87: 1085, 1992; Cancer 70: 1266, 1992; Ann. Int. Med. 126: 811, 1997; Am. J. Med. 106: 1A: 7S, 1999; South. Med. J. 92: 258, 1999; JAMA 289: 1288 & 1297, 2003; flex-sig: South. Med. J. 84: 72, 1991; maybe a flex-sig every 3-5 years over age 50 is a good idea Lancet 341: 736, 1993; Hosp. Pract. 29(8): 25, Aug. 15, 1994; the nurse can do it as well as the gastroenterologist can NEJM 330: 183, 1994; I told you that fecal occult blood screening isn't at all sensitive either for polyps or cancers JAMA 269: 1262, 1993).


• take a good look at the skin for melanomas and other cancers and pre-cancers, especially if there's a history of sun-exposure or some other risk factor (but you should look at everybody: Cancer 75(S2): 674, 1995);


• listen for a carotid bruit, maybe get a Doppler for those old folks;


• do a cheap hearing test;


• check visual acuity (even on kids, Doc: Pediatrics 89: 834, 1992);


• quick glaucoma test if you're an eye specialist;


• ultrasound women for ovarian cancer (probably not, but still a current topic, see Br. Med. J. 299: 1363, 1989; negative appraisal Ann. Int. Med. 121: 124, 1994; negative NIH consensus statement JAMA 273: 481, 1995; more Postgrad. Med. 102: 112, 1997)


• ultrasound older (>50 yr.) people for abdominal aortic aneurysms (well, this was an idea a while back; see Surg. Clin. N.A. 69: 713, 1989)


• serum (or finger-stick; Prev. Med. 20: 364, 1991; Am. J. Card. 65: 6, 1992) cholesterol, starting in teenaged years or before (kids at risk are those with a parent who had coronary disease before age 60, or a known cholesterol problem: Pediatrics 88: 269, 1991, Pediatrics 94: 296, 1994); repeated occasionally to age 75 (older recommendation: every 5 years to age 50); consider fractionating into LDL-c/HDL-c (much-ballyhooed, but not all that useful in the absence of a serious problem; some people check everybody's ratio); despite predictions, apolipoprotein B assay did not replace routine cholesterol screening in the 1990's (Can. J. Card. 8: 133, 1992). NIH consensus document on cholesterol testing (JAMA 269: 505, 1993); review of "diagnosis and treatment of lipid disorders": Med. Clin. N.A. 78: 247, 1994.


• chemical urinalysis (yearly, maybe; there's a case to be made for screening folks over 60 for bacteria, pregnant women for proteinuria, and teens: Postgrad. Med. 100: 173, July 1996)


• syphilis test (old recommendation: every five years to age 50; routine screening seems unpopular today);


• serum iron and iron binding capacity in a man once in a while as a young adult (one of the most cost-effective things you can do: Arch. Int. Med. 154: 769, 1994; Gastroent. 107: 453, 1994; Postgrad. Med. 102: 83, Dec. 1997; Ann. Int. Med. 129: 925 & 962 & 971 & 980, 1998; Postgrad. Med. 102: 83, 1997. not a USPSTF recommendation, which was dim).


• lipoprotein Lp(a) screen, once, for anybody with a family history of unexplained coronary disease (Arch. Int. Med. 157: 1170, 1997).
• serum homocysteine? stay tuned
• hepatitis C screen for swingers and needle users (about 1/3 of hepatitis C positive people have normal SGOT)
• screen for celiac sprue (?!; anti-endomysial or anti-gliadin antibodies: Am. Fam. Phys. 57: 1023, 1998). Since one white person in 300 has it, and it'll make you horribly sick, perhaps this is worth doing occasionally on healthy people, or anybody who's got vague symptoms.
• serum CA-125 screen for women to detect cancer of the ovary (maybe; current subject, Ob. Gyn. 80(3 pt. 1): 396, 1992; Am. J. Ob. Gyn. 165: 7, 1991; Cancer 68: 458, 1991)


• Helicobacter screen for anybody with a family history of this or of stomach cancer Am. J. Med. 106: 222, 1999)


• serum prostate-specific antigen for older men

  (maybe, see above, also J. Urol. 148: 326, 1992; Cancer 69: 1201, 1992; Urology 38: 216, 1991; Cancer 74: 1615 & 1640, 1994; consider using 3 or 4 ng/mL as cutoff; managed care will have to decide what's the most efficient policy: Cancer 71(S3): 981, 1993);

• screen one-year-olds and two-year-olds, and maybe older kids, for lead poisoning



Screening is now in place; you should screen kids if they've ever lived in a house built before 1960, or if you required to do so for reasons of political correctness; remember there's plenty of lead even outside the urban slums: Pediatrics 94: 59, 1994, but the yield in non-slum kids will be very low; remember parents lie about the age of their house (Arch. Ped. Lab. Med. 157: 584, 2003). There is no consensus about when to treat, but it's still good that screening is taking place.

• serum super-sensitive hTSH (or some other thyroid test) to check for pre-clinical thyroid disease (maybe; remember thyroid disease is common, insidious, and devastating; see Clin. Endocrin. 32: 185, 1990; maybe do it every 5 years JAMA 276: 285, 1996);


• blood glucose only if there's some symptom to suggest diabetes (what symptom doesn't?, see for example J. Fam. Pract. 33: 155, 1991);


• EKG (once, baseline, maybe);


• TB skin test unless known positive (occasionally, maybe, if there's lots of TB in the community; older recommendation was every ten years);


• chlamydia screen for young folks (huh?! which ones?) New DNA technology makes it possible to spot chlamydia and treat with one-dose azithromycin, benefitting the community, and mass-screening is now coming into fashion (J. Inf. Dis. 179-S2: S-380, 1999)

Note that this list does NOT include a whole lot of blood work! There are additional interventions (including lab tests) that should be routine during pregnancy, and things to do for children.

My choice for "biggest waste of everybody's time and money"... a sputum cytology "to screen for lung cancer" (J. Clin. Path. 50: 566, 1997). Runner-up: Sed rates "to screen for disease" in healthy people (Postgrad. Med. 103: 257, 1998).

When we figure out how to treat the thrombophilias, we may screen everybody; right now, there's probably no reason to (Ann. Int. Med. 127: 895, 1997).

Today, almost all adult patients who present diagnostic problems (or who somebody wants "screened") receive an automated chemical profile (a panel of colorimetric assays on the serum done by a machine). These are run every morning (maybe not on weekends). The best-known of the machines include the Technicon SMAC, Eastman-Kodak's Ektachem 400, and the Hitachi 7377.

Tests on the profile will usually include:

Total protein
Albumin
Calcium
Inorganic Phosphate
Cholesterol
Uric Acid
Creatinine
Bilirubin
Alkaline Phosphatase (Alk Phos)
Lactate Dehydrogenase (LD, LDH)
Aspartate Aminotransferase (AST, SGOT, GOT)

And the profile at your hospital could also include one or more of these:

Triglycerides
Iron
Total Iron Binding Capacity (TIBC)
Unsaturated Iron Binding Capacity (UIBC)
Direct Bilirubin
Albumin-Globulin Ratio (A/G ratio)
Alanine Aminotransferase (ALT, SGPT, GPT)
Gamma-Glutamyl Transferase (GGT, GGTP, "drunk screen")
Magnesium
Amylase
Creatine Kinase
Glucose
Urea ("Blood Urea Nitrogen", BUN)

"Electrolytes": sodium, potassium, chloride, and "total carbon dioxide content." (These are usually run on a different machine that stays up all night.)

The automated chemical profile provides much information at less cost than any individual test. The machine itself costs around $200,000 to purchase and install. Doing each profile, however, costs pennies in reagents, and a dollar or less overall. (For big batches, one 1983 estimate placed the actual cost to the lab at 17 per profile.)

In the recent past, through creative accounting, the lab would send a bill of $20 or so for the profile. (The clinician in private practice could charge the patient $150 or more for interpreting the profile.) The automated chemistry profile machine has been one of the big money-makers for most hospitals, and it has paid for other services that lose money. And, of course, reducing the ordering of available tests does not result in a linear decrease in the true cost of health care. See Hum. Pathol. 15: 499, 1984; largely of historical interest now.

The lab has other machines to do individual tests on an emergency basis. These typically involve ready-made packages of reagents which cost a few dollars each.

Indications for an automated chemistry profile are still "controversial" because of the "expense" and for other reasons. Whenever there is reason to suspect serious disease (even in such a common situation as mild high blood pressure), and before general anesthesia, a chemistry profile is usually ordered. ("Before general anesthesia" a history and physical is probably sufficient: Mayo Clin. Proc. 72: 505, 1997). Serial profiles may be used to follow the course of a disease and its response to treatment. (A profile every few days is plenty.) Chemistry profiles may be ordered on asymptomatic people, nowadays particularly to screen for cholesterol problems. (All about cholesterol testing: Arch. Path. Lab. Med. 112: 387, 1988). The three most common serious diseases detected by "screening everybody" are (1) hyperparathyroidism, (2) hypercholesterolemia, and (3) alcoholic liver disease. If you screen a lot of older people, you'll also pick up some cases of (4) adult onset diabetes, though these people generally have at least some complaint that warrants screening (obesity, neuropathy, vascular disease, etc., etc.) You'll also pick up a certain number of patients with (5) unexplained elevated liver enzymes ("transaminasitis") which is probably worth working up (most are just drinkers or folks taking ibuprofen or folks with mild NASH/Syndrome X; but you might find something treatable like hepatitis B, hepatitis C, hemochromatosis, autoimmune hepatitis, or Wilson's). It is notoriously impossible to fully rule out Wilson's using any of the common labs (urinary copper, serum ceruloplasmin): Gastroenterology 113: 350, 1997. I feel that these alone justify widespread screening of "asymptomatic adults" by chemical profiles. Not everybody agrees. Pro: J. Gen. Int. Med. 7: 393, 1992; Am. J. Med. 94: 141, 1993 ("$12,000 to find a serious, treatable disease"). Con: Br. J. Gen. Pract. 41: 496, 1991.

The two most common trivial findings are (6) minor permanent elevations of total bilirubin (i.e., "Gilbert's", less often "thal-minor"), and (7) low serum albumin in the chronically sick (why?), which is usually accompanied by low total serum calcium (why?). If you are screening healthy folk, you'll also find (8) plenty of athletes and people doing hard physical labor, who will have elevated creatine kinase. The great drawbacks of mass screening are that data from tests on people without signs or symptoms of systemic disease generates confusion and additional expense to the hospital, and a certain perverse incentive.

* Think. Today's health care plans have a good reason to delay the detection of disease, even if this means the patient will go through a period of sickness, suffering, and disability. If you spend a dollar for a serum calcium screen and you detect a parathyroid adenoma, taking it out will cost the plan big money today. If you let the patient become depressed, have a few years of feeling terrible, and pass a few kidney stones "before screening is justified", he or she is likely to be under the care of a different "provider", and somebody else pays. Get it?

My list, a little longer than the NIH's dementia workup. A psychiatrist (or any other physician) should diagnose psychiatric disease without a physical exam and some labs only if the life-history is absolutely classic for some familiar entity. Remember that fibromyalgia is diagnosed on the history and physical exam.

• chest X-ray (paraneoplastic neurologic effects of oat cell carcinoma are infamous)


• CBC and UA


• chemical profile (notably, how's that calcium? that GGT? those transaminases?)


• syphilis screen


• thyroid screen


• porphyria screen


• serum B₁₂


• serum folate


• f-ANA


• anti-Ro / anti-La (on the "neuro" service see Am. J. Med 90: 479, 1991)


• drug screen


• consider Addison, Cushing, Wilson, pheochromocytoma, Lyme screens


• consider monitored fast to check for fasting hypoglycemia


• consider assays for lead, mercury, other bad metals


• consider cytokine assays or other tests of immune function to detect bone-fide "chronic fatigue syndrome"


• check needle users and swingers for HBV, HCV, HIV

If you have ordered blood tests, the quality of the results depends on your patient and your specimen.

It's your call as to whether your patient needs to fast long enough for the chylomicrons to be cleared from the blood and for lipid levels to return to baseline (i.e., overnight). Otherwise, the "lipemic" blood will cause inaccuracies in most photometric tests. (This limits the usefulness of routine screenings on elective afternoon admissions.) Some labs notice (and tell you about) a "lipemic" serum, others will miss it. High serum triglyceride levels (over 1000 mg/dL) will increase the following values enough to make a difference clinically: hemoglobin, MCH, MCHC; albumin, total protein, fibrinogen, hemoglobin A1c; transaminase enzymes (ALT, AST); bilirubin, calcium, creatinine, glucose, iron, urea. These values will be decreased: amylase, lipase. (The latter's a big deal if you think you patient has hyperlipidemia type I, or "melted strawberry ice cream blood disease".)

If the patient comes to the lab with a full stomach, of course, you may see true elevated glucose (slightly) and triglycerides (up to several hundred mg/dL) and decreased bilirubin (if anybody cares). If the value for the serum iron is critical to your investigations, your specimen should be fasting, and the patient should have taken no iron supplements for a few days. Serum alkaline phosphatase is increased while food is in the small bowel in patients with blood groups O and B who are secretors. If the patient has been on "clear liquids" for a few days, bilirubin may go up. You can figure out for yourself the changes seen in dehydration, ketosis, and protein depletion -- you'll become very familiar with these on rotations.

For routine screening (i.e., you're looking for anemia, early kidney troubles, calcium, transaminases, way-out-of-line cholesterol or glucose) fasting isn't critical. Patients may be unwilling to return in the morning just for bloodwork, so you're usually more likely to get what you really need if you order the labs on the spot.

If your patient is anxious about the needle stick, it will raise serum glucose and catecholamines and sometimes (if hyperventilation occurs) produces alkalosis.

Using a tourniquet to draw blood is often necessary, but causes errors if it is in place for a long time. If there is much fist-clenching, serum lactic acid goes way up, pH goes down, and serum potassium can increase by 1.0 mEq/L (a big increase). By causing transudation of water from the vessels to the interstitial fluid, the tourniquet can also cause small elevations of calcium, iron, magnesium, and proteins.

A common blunder is drawing a blood sample from a site downstream from an intravenous line. If (for example) it's infusing "D5", which is 5% dextrose (glucose) in water, your patient will appear to have a very-high blood glucose and low everything else (Karen Carpenter).

If the blood specimen undergoes hemolysis as a result of clumsy phlebotomy or on the way to the lab (being placed on top of the radiator, in the freezer, dropped, shaken, sipped, etc., etc.), resulting errors can be very serious. Serum magnesium, phosphate, potassium, and some enzymes (LD, AST, acid phosphatase) are greatly increased. LD-1 may be higher than LD-2 ("flip"). (Serum iron will not be significantly increased.) There will also be minor changes in other results, but the false high potassium can result in lethal errors. Serum will be visibly pink when enough hemolysis has occurred to elevate serum potassium 0.1 mEq/L. However, the lab may not notice this.

When blood clots (remember the automated chemistry profile is run on serum), potassium is released from platelets, elevating the value by around 0.5 mEq/L. This is taken into account in calculating the reference range. If, however, the serum stays in contact with the clot for a long while, it will result in inaccurately high iron, LD, and potassium (leaking, mild hemolysis), and falsely lowered glucose (used as energy source by red cells). This is only a problem if the specimen sits around for several hours. The silicone separators help with this. A specimen left on top of a radiator has extremely high LD, phosphate, and potassium, and zero glucose, and this pattern is familiar to every lab worker.

Tiny clots (inadequate mixing) or cold agglutinins in specimens submitted to hematology will result in an increased measured MCV (and decreased calculated RBC -- "uh oh, instant pernicious anemia").

If your patient has recently had a radioactive isotope scan, the results of radioimmunoassays will probably be inaccurate. Check with the lab.

Even if you manage to avoid any cause of analytic inaccuracy, you must consider your patient's level of activity while interpreting your results. Values obtained in patients who have been supine for a while tend to be lower than in patients who were standing up. (Do you know why?) How much are things diluted? 10%: most proteins and non-protein hormones (the ones that are protein-bound); 5%: cholesterol, triglycerides; 2%: hemoglobin-hematocrit; no change: electrolytes. Also remember: standing up is the cause of orthostatic proteinuria, the commonest reason for excess protein in the urine in outpatient practice.

Childhood: At birth, hemoglobin is around 19 mg/dL and MCV around 105 fL. (Both are probably higher than yours -- why might this be?) A few nucleated reds are normal in a newborn's blood. These parameters drop off to lows of about 11 mg/dL and 78 fL at one year, then gradually increase to adult values by puberty. The absolute counts of white cells vary depending on age. At birth, all normal white cells are increased 2-4x the normal adult value. Bands and segs peak during the first day, then drop off. Segs are lower than adult values during the first year of life, and the count gradually increases over the following years. The other cell types diminish to near-adult values around age 6. Alkaline phosphatase is 2-5x normal (growing bones), dropping to adult levels when bone growth stops after puberty. Cholesterol levels in healthy children are of course low. Alpha-1 protease inhibitor levels are usually low ("uh-oh, instant emphysema"), and angiotensin converting enzyme levels are generally high ("uh-oh, instant sarcoidosis").

Menstruation: Remember that the platelet count may drop by 50% during a normal period.

Pregnancy: "Anemia of pregnancy" is usual, but hemoglobin should not go below 10 gm/dL. There is always hemodilution, but more severe anemia suggests iron deficiency and/or folate deficiency. Glucose tolerance decreases. Most serum proteins change as in the acute phase reaction ("stress of being pregnant"). However, serum transferrin increases (and serum iron decreases, of course). Serum alkaline phosphatase increases as the placenta grows (2-3x normal). Granulocytosis develops before and during labor.

Old age: Alkaline phosphatase increases, often to several times the "upper limit of normal", without any disease. Cholesterol levels average higher than in young adults. The other "routine" chemistry and hematology values do not change significantly in "healthy old age". Various antibodies, including rheumatoid factor, anti-nuclear antibodies, and some syphilis antibodies, may appear without there being disease.

Runners and other trained athletes show a characteristic pattern of changes. (See Mayo Clin. Proc. 55: 113, 1980) (1) increased plasma volume with decreased Hgb, Hct ("jogger's anemia", still a little bit mysterious); (2) increased reticulocytes (hemolysis due to pavement pounding); (3) increased high-density lipoprotein ("HDL-c", the "good" cholesterol), and decreased low-density lipoprotein ("LDL-c", the "bad" cholesterol); (4) increased creatine kinase (CK, CPK) with up to 15% MB fraction (the "myocardial damage" fraction); (5) Increased AST (hepatic ischemia, after marathons. See JAMA 252: 626, 1984). No labs are useful in screening for sports participation (J. Fam. Pract. 52: 127, 2003; interesting paper which matches the way I've always done and taught it).

Estrogen users (oral contraceptive pill, etc.) exhibit a characteristic group of changes: (1) increased thyroxine, decreased T₃RU (increased thyroid-binding globulin); (2) increased cortisol (because of increased cortisol-binding globulin); (3) decreased antithrombin III (remember these people may get thrombosis); (4) increased ceruloplasmin, sometimes increased GGTP; (5) increased triglycerides (maybe)

Better to sleep with a sober cannibal than with a drunken Christian. -- Melville, Moby Dick

Alcoholics and other heavy drinkers present many problems. Every clinician would like to have a simple test to identify problem drinkers accurately. You know the CAGE questions:

C: Have you tried to Cut down drinking?

A: Are you Annoyed by others' concern about your drinking?

G: Have you felt Guilty about drinking?

E: Have you use alcohol as an Eye-opener?

There are others (Arch. Int. Med. 155: 1726, 1995). To start off, look for tolerance. Anyone who isn't obviously drunk with a blood ethanol of 0.15% drinks often and heavily enough to have a tolerance. The same is true for anyone who is walking (driving, etc.) with a blood ethanol of 0.3% (these people are supposed to be more-or-less dead). In the North, I was taught that people who identify themselves as "moderate drinkers" are all chronic heavy drinkers. A serum AST greater than serum ALT, in an otherwise healthy person, suggests heavy drinking. High MCV, high serum uric acid, low serum phosphate, high triglyceride (after parties, in "type IV" patients), and high alkaline phosphatase all have been examined by hopeful laboratorians as signs of habitual drunkenness. (History of trauma helps too, see Ann. Int. Med. 101: 847, 1984.) The most enthusiasm in recent years has been over gamma-glutamyl transferase (transpeptidase, etc.), elevation of which is common in heavy drinkers (30-80%, often the only abnormality on a profile.) Of course, the same picture can be seen in mild pancreatitis, liver problems (including CHF), phenytoin therapy, estrogen therapy, rheumatoid arthritis, a wide variety of cancer-related problems, and probably others. (See Br. Med. J. 286: 531, 1983.) However, most of the above are easy to rule out, and GGT is fairly well established as a means of following compliance with therapy in many alcoholic treatment programs. (GGT update: Br. Med. J. 302: 388, 1991). Some labs can measure the acetaldehyde-induced hemoglobin fraction (HgbA_1ach), a marker for how much alcohol one has drunk in the last few weeks (Arch. Path. Lab. Med. 116: 924, 1992). Lab testing to spot alcoholism: West. J. Med. 156: 287, 1992. "Carbohydrate-deficient transferrin" (CDT test) is now a (fad?) assay to screen for alcohol abuse and test one's abstinence (Clin. Chem. 39: 866, 2001). It is also likely to be up if is there is already advanced liver disease (Alcoholism 25: 1729, 2001), in women losing weight (Alcohol and Alcoholism 36: 603, 2001), and in many others. * Future tests may include "mitochondrial AST", and other exotica (Alcoholism 16: 82, 1992; Am. J. Med. Sci. 303: 415, 1992, QJM 89: 137, 1996; seems to have fallen by the wayside.).

Diuretic abusers and laxative abusers exhibit decreased serum sodium, potassium, total CO₂, magnesium, and increased chloride (why?). The low potassium can be very unhealthy.

Anorexia nervosa types (including "flippers") show surprisingly normal lab tests. Endocrine testing may reveal diminished (but not absent) pituitary response to stimulation testing, which compounds confusion with Simmonds's disease.

Anabolic steroid users ("breakfast of champions") also exhibit a characteristic group of changes: (1) Increased serum creatinine (increased muscle mass); (2) increased LDL-c cholesterol, and decreased HDL-c cholesterol (see JAMA 252: 507, 1984); (3) increased hemoglobin and hematocrit; (4) increased direct bilirubin (cholestasis), and sometimes increased alkaline phosphatase; (5) decreased thyroxine, increased T₃RU (decreased thyroid-binding globulin); (6) decreased serum phosphate.

Physical examination: Some authorities claim that vigorous prostate massage causes mild (2x normal) increase in serum acid phosphatase and/or prostate-specific antigen for a day or so. (Why?) The changes seem real, but aren't going to mislead any clinical decision-maker (JAMA 267: 2227, 1992).

Intramuscular injections: Remember these will raise serum skeletal muscle enzymes. (Why?)

Drugs: Remember that drugs can affect lab tests in two different ways.

1. The drug may act as an interfering substance in the assay ("methodologic interferences"). For example: (a) vitamin C in high doses causes false negative urine tests for glucose and blood; (b) tetracyclines, being fluorescent, cause false positive fluorescence tests for porphyrins and catecholamines; (c) tolbutamide, erythromycin, and several others cause false increases in transaminases when colorimetric methods are used; (d) bromide ion is measured as chloride by most electrodes. And so forth.

2. The drug may truly change the physiologic parameter being measured. For example, (a) spironolactone decreases serum sodium; (b) glucocorticoids should suppress ACTH production and raise blood glucose; (c) methyldopa often causes antibodies to bind to RBC's, making them "Coombs positive"; (d) procainamide often causes a "lupus-like" syndrome, with positive anti-nuclear antibodies; (e) gold injected for rheumatoid arthritis can cause proteinuria; (f) a wide range of drugs can cause decreased platelets and/or neutrophils. These are not "artifacts", they are real.

Several of the tests on the chemical profile are enzyme assays. When cells are damaged (not necessarily irreversibly), they may lose some of their contents into the bloodstream. Often these contents include the intracellular enzymes that make up the machinery of the cytoplasm. (Or very active cells release extra enzyme). If serum levels of a particular enzyme activity are elevated above normal, it commonly indicates injury to the cells, and the species of enzyme increased in the serum indicate what organ is damaged.

The enzyme activities are assayed rather than the enzyme proteins themselves. (Some newer techniques, such as "prostatic acid phosphatase", measure a specific protein by immunologic techniques. This does not necessarily make the test more useful clinically.) Enzyme activities are measured in units of activity, defined differently for different enzymes. (The enzymes themselves would probably be measured in ng/dL.) An "International Unit" (U, IU) of an enzyme is the amount that catalyzes the conversion of 1 micromole of substrate per minute under the defined conditions of the test (temperature, pH, substrate concentration, etc.) Speaking of units, an attempt was made to get all clinical labs and journals to switch to the SI (International System) standard and report everything in terms of kilograms, meters, and seconds, beginning July 1, 1988. It was a total failure (NEJM 327: 49, 1992).

If your tube of blood must wait for two or more hours before being assayed, put it on ice.

You will need to learn these enzymes, which are found in several different tissues, because they are commonly measured by the clinical lab.

Alanine transaminase (ALT, SGPT, GPT): liver, skeletal muscle

Alkaline phosphatase (alk phos): bone (osteoblastic activity), intestine (secretors with blood types O or B, or after a fatty meal), liver (especially obstructive-metastatic disease), placenta, some cancers ("Regan" isoenzyme).

The serum levels of three enzymes correlate with hepatic alkaline phosphatase: "5'-nucleotidase" (5'-NT, 5'-N), "gamma-glutamyl transpeptidase" (GGTP, GGT), and "leucine aminopeptidase" (LAP). Checking one or more of these can clarify the origin of an elevated alkaline phosphate.

Amylase (alpha amylase): pancreas, salivary gland, and (to a much lesser extent) many other organs

Aspartate transaminase (AST, SGOT, GOT): heart, liver (especially hepatocellular disease), red cells, skeletal muscle

Lactate dehydrogenase: most organs, especially heart, liver, lung, kidney, red cells, skeletal muscle, many cancers

Isoenzymes are variant forms of enzymes found in different tissues. They do the same things, but they may be composed of different subunits or be entirely different proteins. Knowing which isoenzyme is elevated will often tell which of several tissues has been damaged.

The most famous isoenzyme change is the "LDH 1-2 flip" (LDH-1 higher then LD-2). It indicates necrosis (infarction) of the heart or kidney, or hemolysis. (High LDH-5: Liver, skeletal muscle. LDH-3 high? Think of a pulmonary infarct. All high? Shock or infectious mononucleosis! Why?)

LDH 1: KNOW: Heart, kidney, red cells and their precursors (very sensitive to mild hemolysis during or after sample collection), seminoma
LDH 2:
LDH 3: KNOW: Lung, lymphocytes, platelets
LDH 4: (some acute lymphoblastic leukemias produce a lot of this)
LDH 5: KNOW: Liver, skeletal muscle, neutrophils, prostate

Indications for serum calcium include suspicion of cancer, symptoms or signs of hyperparathyroidism (kidney stones, psychiatric problems, many others), bone diseases, unexplained seizures (especially in babies) or coma, after thyroid or parathyroid surgery, or in other patients at special risk (see below).

The lab measures total calcium, i.e., ionized plus un-ionized. Only about half the calcium in the serum is the physiologically active ionized calcium. The rest is bound to albumin and other proteins, and protons compete for binding sites. The fraction of serum calcium that is ionized varies with pH, because protons compete with calcium to bind to albumin. A decrease of 0.1 pH unit increases the ionized fraction of calcium by about 2%, so the effect is not very marked. However, the fraction of serum calcium that is ionized also varies with the serum albumin level. A decrease of 1 gm/dL in albumin decreases both un-ionized and total calcium by about 0.8 mg/dL. This is important.

The lab does not report ionized calcium (even though that is what you are really worried about) because measuring this requires special techniques for blood collection, a calcium-specific electrode, and other resources. Around normal pH, several formulas have been devised to estimate ionized calcium. For example:

* ionized Ca++ = (97 - (8 x albumin) - (2 x globulin))/100

Or, at a glance:

Ionized Ca++ (mg/dL) ~ Total Ca (mg/dL) - albumin (g/dL)

Many interesting laboratory procedures have been devised to distinguish hyperparathyroidism from other causes of hypercalcemia. You will learn about electrolyte equations, serum parathyroid levels, urinary cyclic AMP, tubular resorption of phosphate ("TRP"), etc., etc., on rotations.

Increased serum magnesium is usually due to renal failure (or to hemolysis of the specimen). Decreased serum magnesium is a subject of considerable current interest, and is thought to be widespread among the sick.

This will get you started....

Iron deficiency: Low serum iron (Fe), high serum iron-binding capacity (TIBC), near-zero ferritin

"Anemia of chronic disease": Low serum iron, low serum iron-binding capacity, normal or high serum ferritin

Iron overload: High serum iron, Fe/TIBC>60% or so, high serum ferritin

Making sense out of this:

1. In iron deficiency, the capacity of the serum to bind iron increases "in the hopes of snarfing up more iron".

2. In anemia of chronic disease (i.e., prolonged interleukin 1 release making it difficult to transfer iron from the bone marrow macrophages to the normoblasts), interleukin effect also causes lowered TIBC (like albumin, a protein which goes down in the acute phase reaction).

3. Transferrin saturation (i.e., Fe/TIBC) of >60% is suspicious for iron overload, and >80% is extremely suspicious. Remember that transferrin levels drop during the acute phase reaction.

4. Of course, if your patient just took a handful of "multiple vitamins with iron" prior to visiting you, interpretation is clouded....

5. Serum ferritin shouldn't be more than 300 ng/mL in a man, or 200 ng/dL in a woman. Levels closely parallel total body iron stores; the fact that serum ferritin rises somewhat in the acute phase reaction shouldn't cause trouble.

Caveat: Many allied health professionals use "iron" as a synonym for blood hemoglobin content. I've given up arguing.... As noted, if you're really concerned about iron, draw a fasting specimen.

*There's a circadian rhythm to serum iron but it's slight and shouldn't matter: Am. J. Clin. Path. 117: 802, 2002.

You'll learn the protocol for the workup of suspected myocardial infarct on rotations. Criteria underwent major changes in 2001 (Am. Heart J. 144: 957, 2001). Worth remembering:

Troponin I or troponin T is your best enzyme because of tissue specificity. Draw your levels every six hours (NEJM 330: 670, 1994). CK, SGOT, and LDH are less specific and offer no advantages, so are not recommended.

Creatinine kinase levels rise following heavy exercise and can stay way up for days. A few % MB band is commonplace, especially after a heavy workout or heavy-duty muscle injury (i.e., surgery). Intramuscular injections can bring a couch potato's creatine kinase to the upper limit of normal. An MB band of maybe 15% of the total CK is pretty suggestive of myocardial injury. The band should disappear 18 hours after heart surgery.

Elevation of the LD1 isoenzyme ("1-2 flip") supports your diagnosis of myocardial infarct (or indicates hemolysis in patient and/or blood tube, and/or renal injury). CK rises and falls, typically in a day or so; the MB is even more ephemeral. SGOT rises and falls more slowly (1-3 days), while LDH stays around even longer

Don't expect to see elevated CK-BB in brain injury (it may not cross the blood-brain barrier, or be present in large enough amounts, or whatever).

Of course, you will continue to look for the requisite Q-waves and/or ST-segment elevation on EGK, decreased wall motion on echocardiography, and hypoperfusion on scan. Remember the latter two are only about 50% specific.

In the old days "Serology" meant a syphilis test. Today, the serology lab does most of the procedures involving immunology. There is no need for you to know about the technical aspects of these tests.

The titer of a substance measured in the serology lab is the maximum dilution (of a series of dilutions) at which the substance can be detected. Thus a titer of 1:2 or 1:10 is a "low titer" and indicates that not very much of the substance is present. And a titer of 1:128000 is probably a "high titer". Depending on what you are measuring, a titer of 1:100 might be "high" or "low".

"A significant rise in titer" suggests a recent infectious disease. (Why?) "Significant" is usually considered to be a "fourfold rise". If a titer rises from 1:16 to 1:64 during an episode of acute illness, or when a titer rises from 1:10000 to 1:80000, the patient probably had the acute disease to match.

Effusions are ultra-filtrates of plasma in the pleural, pericardial, or peritoneal cavities. These are commonly encountered in sick people. You may remove fluid which has accumulated in a cavity for diagnostic purposes, or to treat the patient (pleural effusions may restrict lung expansion, pericardial effusions may cause tamponade, etc.) Here is one recommended way to submit a sample of the effusion to the lab for study:

• one EDTA tube to hematology for a "CBC and diff"
• one heparin tube to microbiology for stains and cultures (aerobes, anaerobes, acid-fast bacilli, fungi)
• one heparin tube to cytology (check for cancer cells, etc.; remember that "negative cytology" doesn't mean there's no cancer in the body cavity, only that the cells aren't being shed into the effusion!)
• one plain tube to chemistry. Select such tests as....

LD (LDH; compare with serum values to determine whether the effusion is a transudate or exudate);

glucose (low in rheumatoid and septic effusions);

triglycerides and cholesterol (milky effusions);

amylase (acute pancreatitis; remember this can also cause pleural effusions);

* bile (bile peritonitis, as from a ruptured gallbladder);

* hyaluronic acid (reference lab procedure which may help make the diagnosis of mesothelioma).

Transudates generally are due to congestive heart failure, hepatic cirrhosis, or the nephrotic syndrome. These effusions typically contain less than 2.5 gm/dL protein, show specific gravity less than 1.015, effusion LD:serum LD less than 0.6. These characteristics must all be present, or the fluid is not a transudate. Transudates fail to clot, and contain fewer than 10 cells/cu mcL (lymphs, mesothelial cells). (These can be true of exudates, too.) If the protein and LD values indicate a transudate, no further tests on the fluid are indicated. (You may discard the other tubes.)

Exudates should make you think of cancer or infection (TB, pneumonia, etc.) Less often, exudates are due to trauma, pancreatitis, rheumatoid arthritis, systemic lupus, infarcts (pulmonary, myocardial, etc.), trauma. Exudates contain more protein and LD than transudates, and may even clot. The type of cells in an exudate can help tell its origin. Many polys (i.e., a "purulent exudate") of course suggests bacterial infection. Bloody effusions suggest cancer, infarction, or iatrogenic or other trauma. Perhaps 50% of malignant effusions contain cancer cells. Lymphocytes and mesothelial cells are quite nonspecific.

Milky effusions always raise the concern that the thoracic duct has been damaged (cancer, trauma), producing a true "chylous" effusion. A "chylous" effusion will of course be rich in chylomicrons, which give the effusion a high triglyceride content, may layer at the top of the collection tube, etc. Much more common is a "pseudochylous" milky effusion, the result of cell degeneration in any chronic effusion. Color, turbidity, and odor are variable; cholesterol content is high and cholesterol crystals often impart a "gold paint" appearance to these chronic effusions.

To see whether the person has malabsorption, we recommend a fecal fat stain. The lab will put some oil-red O on a smear of stool, and look for fat blobbies. If there are none of these, and the patient eats a reasonably normal American diet of greaseburgers and fries, you don't have malabsorption. If these are present, your patient has malabsorption or maybe a plugged bile duct (but you'd have known the latter).

To distinguish pancreatic insufficiency ("maldigesetion", for purists) from intestinal insufficiency ("true malabsorption"), do a d-xylose test. Administer this simple sugar by mouth, and see if it ends up in the urine. If it does, the intestine works, the malabsorption is pancreatic, and you must figure why (cancer, alcoholism, cystic fibrosis, rarely anything else subtle). If it doesn't, the intestine is out of order, and you ought to do a biopsy.

Some reminders: Most definitive diagnoses of serious diseases like cancer are made by examining tissue -- cytologies, biopsies or larger surgical specimens. You may still be taught to "do all the non-invasive procedures first." These includes blood and urine tests, cultures, x-ray studies, and so forth. (Remember that the single most useful diagnostic procedure is still the history and physical exam.)

When it comes time to get tissue, you may...

(1) obtain (or help obtain) cells for cytologic study by the pathologist and assistants (by "noninvasive" Pap smear of the cervix or other accessible sites, or by "invasive" needle aspiration). Many pathologists like to do their own fine-needle aspirations; you can learn the technique, also.

(2) obtain (or help obtain) a tiny piece of tissue for the pathologist to study (a biopsy, the opposite of necropsy/autopsy). Clinicians usually do the biopsies.

(3) assist at surgical removal of tissue.

(4) obtain an autopsy permit. (* How to do an autopsy properly: Arch. Path. Lab. Med. 118: 19, 1994).

Everything that comes out of a patient at surgery goes to the pathologist. (Future surgeons -- remember that the tissue-formalin ratio should be 1:10 or less.) If a diagnosis is needed at time of surgery, the pathologist will perform a frozen section on some tissue. These are very accurate, though the pathologist may defer final diagnosis, especially if the lesion is a papillary tumor or a possible malignant lymphoma. The pathologist will probably be able to tell you something when the slides come out, usually on the weekday afternoon after you obtain the tissue.

We pathologists are pretty good; the latest study in which all diagnoses were reviewed by a second pathologist found an error rate of 0.26% (Am. J. Surg. Path. 17: 1190, 1993; these are errors that might make a difference clinically; the cost of a second pathologist would be $2700 per error avoided.) Pathologists show about 20% of their cases to friends in-house, and this helps: Am. J. Clin. Path. 117: 751, 2002.

* Future clinicians: Keep your biopsy instruments sharp so as not to introduce "crush artifact"!

* Telepathology, the process of turning glass slides into digital images and sending them for diagnosis to one or more pathologists at remote sites, is still much slower and significantly less accurate than conventional microscopy (Arch. Derm. 139: 637, 2003).

Here are some unusual labs and why you might want to get them!

Pyroglutamic acid (5-oxoproline)(urine)... Glutathione synthetase deficiency (metabolic acidosis, mental retardation)

Trimethylamine (urine)... Trimethylamine oxide synthetase (stale fish disease)

Analytical sensitivity is the ability to detect the presence of a given substance, expressed as the least concentration detectable or measurable by the test. Modern laboratory tests include very sensitive procedures. Serum vitamin B₁₂ levels are routinely measured in picograms. The most sensitive pregnancy tests detect the hCG produced by trophoblastic cells within a few days following implantation. The polymerase chain reaction, which detects the presence or absence of a single copy of a gene, is now coming to the hospital bench. Some procedures to detect "abnormal substances" (for example, some super-duper CK-MB isoenzyme tests) are not used by the lab because they are too sensitive, and detect traces of the "abnormal substance" in most normals.

Analytical specificity is the ability to detect a single specific substance and no others. Most tests in the hospital lab are not absolutely specific, though this is not usually a problem for clinicians. For example, bromide ion (if present) will be measured as chloride ion by routine "electrolyte studies." Acid phosphatase measurements always include a contribution from alkaline phosphatase. Some procedures to measure AST and creatinine levels also measure acetoacetic acid, so AST and creatinine values will appear falsely elevated in starvation and diabetic ketoacids. Lipemia (cloudy fat in the serum) may falsely increase (or decrease) the results of many common chemical tests. And so forth. You will learn about most of the important problems during this course.

Accuracy is closeness with which an obtained result agrees with a known accepted true value. In addition to routine quality control, most clinical labs report their results on samples provided by outside agencies (such as the College of American Pathologists) to check on their own accuracy. See Arch. Pathol. Lab. Med. 109: 709 & 1066, 1985.

Precision is the reproducibility of test results. Serum electrolyte studies typically vary by only 1% or even less. (This is good, as "'lytes" need to be tightly regulated.) Most other chemical tests may vary by around 5%. Measurements of enzyme activities may vary 10% or even more. Again, this is not usually a problem for clinicians.

The following concepts come up continually in discussions of lab data, and are favorite licensure exam items. Lab data interpretation usually starts with comparing your patient to whatever it "normal." The reference range ("normal range"): range of values in which a healthy person's test result is "expected" to fall. This is usually the mean + 2 S.D. for "healthy" members of the population to which the patient belongs. Notice that 5% of "healthy" people arbitrarily fall just outside the normal range. The "population" is generally healthy adults, though for some tests, there are separate "normal ranges" for different ages, races, sexes, and so forth.

Many other models for defining the reference range exist, including some extremely cumbersome techniques where the distribution is non-Gaussian. Medical students who do not have the good fortune to attend KCUMB work hard to memorize "normal ranges" for the common tests, sometimes discovering that they vary greatly from hospital to hospital because of different techniques or even different units of measurement. You may encounter the Centinormal system, in which the "upper limit of normal" for each clinical chemical test is set at 100.

A more satisfactory approach to the problem of "normal ranges" could be to establish certain abnormal values as decision levels ("action levels", etc.), depending on whether the clinician should actually do anything about the "abnormal" result. The most familiar example is the recommendation that anyone with a cholesterol over 200 mg/dL or LDL-c over 160 mg/dL or HDL-c under 35 mg/dL should take action to improve things.

However you do it, you will decide that each diagnostic procedure's results are either positive (suggesting a disease is present) or negative (suggesting a disease is absent). Unless the diagnostic procedure you have used is the absolute reference criterion for a particular disease (that's usually an autopsy), your results are also either true or false as indicators of disease. (Please use common sense. Don't just tell your patient, "Your lab results are all negative." The patient might not understand.)

Diagnostic sensitivity ("Bayesian sensitivity") is the percentage of positive results in patients with a particular disease.

Diagnostic Sensitivity=100 X (TP)/(TP+FN)

Diagnostic specificity ("Bayesian specificity") is the percentage of negative results in patients without a particular disease.

Diagnostic Specificity=100 x (TN)/(TN+FP)

Sensitive tests are best for the diagnosis of treatable diseases: bacterial infections, early cancer, phenylketonuria. You want a very sensitive test when the benefits of detecting the disease are great (curing it, preventing new cases, etc).

Specific tests are best for the diagnosis of non-treatable diseases: chronic neurologic disease, disseminated cancer, etc. You want a very specific test when the risks of a wrong diagnosis are great (getting very upset, losing your insurance, getting cancer chemotherapy, etc.)

These are appalling over-generalizations, but there is always a tradeoff between sensitivity and specificity. If a new test is both more sensitive and more specific than an old test (and not much more expensive), it replaces it. Otherwise, whether we are a pathologist setting a "reference range" or "decision level", or a clinician ordering a lab test, we must remember that sensitive tests lack specificity and specific tests lack sensitivity. Clinicians commonly order the sensitive tests first, followed by the specific ones.

Predictive value: the percent chance that a result correctly identifies the patient as diseased or non-diseased.

Predictive Value of a Positive Result = 100 x (TP)/(TP+FP)

Predictive Value of a Negative Result = 100 x (TN)/(TN+FN)

Diagnositc Accuracy ("Diagnostic Efficiency"): 100 x (TP+TN)/(TP+FP+TN+FN)

Prevalence -- How would you calculate how many people are sick?

Prevalence: (TP+FN)/(TP+FP+TN+FN)

Prevalence may be expressed in different units. Contrast incidence: the fraction of new cases of the disease in a population over a given time.

Prevalence = Incidence x Average Duration

* Bayes' Theorem.

PV of a + result = ((sens)x(prev))/((sens x prev.)+((1-spec)x(1-prev)))

Working problems:

If someone gives you a prevalence, sensitivity, and specificity, here is an approach to finding predictive values.

Consider one million people.

1000000 x prevalence= number sick

Number sick x sensitivity = TP

Number sick - TP = FN

1000000 - number sick = number healthy

Number healthy x specificity = TN

Number healthy - TN = FP

Take it from there. If you need, you can work backwards, too.

TO THINK ABOUT:

1. "Jargon is not insight." What are some reasons people use jargon?

2. What might be some reasons that US physicians order more lab tests than our British counterparts? (See JAMA 252: 1723, 1984 and Lancet 1: 1278, 1984.)

3. Why might it be easier to talk about lab test results ("the numbers") than about changes in the physical exam ("the protoplasm")?

4. If twelve tests are run on a panel, and a healthy person has only .95 chance of being in "normal range", what are this person's chances of having at least one abnormal test result?

5. What are some reasons for doing all the non-invasive tests first?

6. Why are enzyme activities, rather than the concentrations of the enzymes themselves, usually measured?

7. What are some reasons that clinicians like to look at the glass slides on their own biopsy and surgical specimens?

8. Will a biopsy of a colon cancer tell you the grade? The stage?

9. What are some problems with setting a "normal range" for serum cholesterol in the U.S.? (See Arch. Path. Lab. Med. 112: 387, 1988).

10. What were some of the problems when mass screenings of Afro-Americans for sickle cell disease were introduced in the late 1960's?

11. Suppose ankylosing spondylitis is an unpreventable, easily-diagnosed, untreatable, mildly disabling disease, that only the 4% of people who are HLA-B27 positive are at risk, and that the risk even for them is 1 in 100. No one suggests that being HLA-B27 positive is a reason to forego parenthood. It is proposed to screen visitors to the shopping mall using a new, fifty cent, highly accurate test for HLA-B27. What do you think of this suggestion?

12. You want to order a routine CBC and chemical profile to screen an asymptomatic adult in your office. Can you draw the blood now, or will you make the patient go to the lab tomorrow morning before breakfast?

13. What are some reasons that the lab does not publish "sensitivity" and "specificity" data for each test?

14. A disease has a prevalence of 1 in 100,000 people. A screening test for this disease involves flipping a coin, with "heads" a positive test. What is the predictive value of a negative result?

15. It is proposed to use a particular test to screen for a certain disease. The prevalence of the disease in the population being screened is 1%. The sensitivity and specificity of the test are both 95%. What is the predictive value of a positive result for this screening test? Does it change if the prevalence is 10%? If the prevalence is 50%?

Lash's Bitters
Pittsburgh Pathology Cases