tion, laboratory tests) required by the protocol Management of protocol; (2) the protocol could assist the nurse in making correct diagnostic , Urinary Frequency, assumptions, recommending appropri­ ate therapy, and referring the complex and high-risk patients to the physician; and Vaginal Discharge and (3) use of the protocol by the nurse could save physician time. Sheldon Greenfield, MD, Gerald Friedland, MD, Sally Scifers, RIM, Materials and Methods Arthur Rhodes, MD, W. L. Black, ScD, and Anthony L. Komaroff, MD Description of the Protocol - This Los Angeles, California, Boston, Massachusetts, and Lexington, Massachusetts protocol provides a rational, logical, and medically sound approach to the diagnosis and treatment of women with the symptoms of frequent or A protocol to be administered by nurses for the management of dys­ painful , vaginal discharge, uria, frequent urination, and vaginal discharge was validated. In a ran­ and vaginal irritation. It directs the domized, controlled trial, 146 women were seen by both nurse and user to collect relevant data (history, physical examination, and laboratory physician and then assigned to either the nurse-protocol treatment tests), and then it guides the user plan or the physician treatment plan. The clinical data collected by through a sequence of decisions to a the nurse showed no important differences from the physicians’ data. diagnosis of uncomplicated urinary The protocol recommended that 89 percent of the patients be sent tract ; urethritis; monilial, tri- home without seeing the physician. The physicians agreed with the chomonal, or nonspecific ; or protocol-recommended disposition in all but two cases. All patients a combination of these. For each of these diagnoses the protocol specifies with complications were appropriately referred to the physician. In therapy and disposition. The protocol follow-up, more than 95 percent of both groups reported sympto­ identifies pathologic entities, such as matic improvement, and repeat urine cultures were negative. We con- gonorrhea, , and hypertension, dude that the protocol can be accurately administered, makes sound that may be related to the presenting complaints either as causes, results, or recommendations, is safe, and efficiently saves physician time. complicating factors. If the nurse sus­ pects these more complex pathologic entities, the protocol directs referral of Several groups in the United States cian time. Protocols assure highest the patient to a physician. are testing protocols (also called clin­ quality care both by guiding perfor­ The protocol is a single-page form ical algorithms) for common clinical mance and by providing a system to that combines a data collection form problems. The protocol is a form that audit effectiveness. with the decision logic pattern neces­ combines a data collection section Komaroff and associates1 devel­ sary to direct data gathering, diagnosis, with a decision logic pattern to direct oped and used protocols for the return therapy, and disposition appropriate the practitioner in history taking, diag­ visits of patients with diabetes mellitus to the particular patient. nosis, therapy, and disposition appro­ and hypertension. Sox, Sox, and The medical judgments in the pro­ priate to a particular patient. Protocols Thompson2 have extensive experience tocol were subjected to consultant re­ can be used as aids to train and guide with the computerized audit of clinical view. The protocol is described here nurses, nurse-practitioners, and physi­ algorithms for acute minor illnesses as and shown diagrammatically in Figure cian’s assistants, and to audit diagnos­ part of a physician’s assistant training 1. If the patient has dysuria or fre­ tic and therapeutic decisions. By facili­ program. We did an uncontrolled vali­ quent urination or both, relevant his­ tating the delegation of clinical- care dation study3 of a protocol for upper tory is obtained, physical examination responsibilities to providers other than respiratory that included and urinalysis are done, and the urine physicians, protocols conserve physi- some measurements of the quality of sample is cultured. If the urinalysis care delivered. We report here a ran­ shows at least 20 or more leukocytes domized, controlled, clinical trial of a or 2+ bacteria per high power field in a protocol for the management of fre­ centrifuged sediment, the patient is From the Ambulatory Care Project, Depart­ quent and painful urination, vaginal treated for .4’5 ments of Medicine and Preventive Medicine, discharge, and vaginal irritation. The If the urinalysis results are negative, CLA School of Medicine, Los Angeles, Ulifornia; the Department of Medicine, performance of a nurse using the pro­ then a pelvic examination is done to eth Israel Hospital, Harvard Medical rule out vaginitis. If vaginitis is not e ool, Boston, Massachusetts; and Massa- tocol was compared with that of a usetrts Institute of Technology, Lincoln group of physicians; process and out­ present the patient is considered to a oratory, Lexington, Massachusetts. Re- come criteria for quality of care and have or urethritis. n ' S , rePr'nts should be addressed to “ - Sheldon G re e n fie ld , D e p a rtm e n t o f Pre­ efficiency were measured. The patient is informed about the con­ ventive Medicine, UCLA School of Medi- dition but is not treated with antibiot­ h«k S An9?les- Calif 90024. This article We tested three hypotheses: (1) the thn, 6en rePr'nteU with permission of au- nurse could accurately collect the clin­ ics until culture results are evalu­ S1.,s and Publisher from Ann Intern Med 81 -452-457, October, 1974. ical data (history, physical examina­ ated.6,7 A patient complaining of ei-

THE JOURNAL OF FAM ILY PRACTICE, VOL. 2, NO. 3, 1975 ther vaginal discharge or irritation of the vulva receives a pelvic examina­ tion, including inspection for gross ab­ normalities and tests for monilia or trichomonas. If neither are detected, the patient is treated for non-specific vaginitis. To check for gonorrhea, the proto­ col directs inspection and palpation of the cervix for purulent discharge and tenderness. If either is present, a Gram’s stain and culture are done. At the UCLA Student Health Service, where the study was conducted, a cul­ ture for gonorrhea is taken routinely with each pelvic examination, regard­ less of the clinical findings. Patients are automatically referred to the physician if they are on a return visit for any of the symptoms listed above, are older than age 45, have had a recent gynecologic procedure, are pregnant, or have diabetes, severe ab­ dominal pain, back pain, incontinence, vomiting, nausea, fever or chills, pro­ teinuria, glycosuria, hypertension, sig­ nificant fever, or any vaginal abnor­ malities. A history of recurrent urinary tract infection, chronic kidney disease, or medications recently administered vaginally are reasons for verbal consul­ tation by the nurse with the physician, after the examination is completed, but the physician need not see the pa­ tient. Urinary tract infections are treated with sulfisoxazole.6 If the patient is allergic to sulfonamide, then tetracy­ cline or ampicillin is used. Nystatin Figure 1. Major logic pathways through the urinary tract infection (UTI) and vaginitis (Mycostatin®, E. R. Squibb and Co., protocol. The protocol specifies what is meant by phrases such as "worrisome symp­ toms" and "toxic." Beside this logic, the protocol contains many other minor path­ Inc.) suppositories are prescribed for ways and logic branch points. the treatment of monilia infection and metronidazole for the treatment of tri­ chomonas infection. Vaginal supposi­ tories of sulfonamide are used in the treatment of nonspecific vaginitis. The accuracy of the initial diagnosis is checked by the culture test results, STUDY DESIGN and treatment is modified if necessary. Patients are advised to return if rash, P h y s ic ia n .Nurse-protocol — ►Physician — fever, chills, or vomiting develop or if Treatm ent the symptoms continue for more than / 1— i— 1 three days with urinary tract infection A ll B lin d Comparison and more than seven days with vagini­ Patients Comparison of tis. (Random ) 0 Df a Outcome x. Collection Study Design — The study con­ i------1 Nu rse-protocol ducted between March 1, and May 31, P h y sic ia n -— ► Nu rse-protocol — * Treatm ent 1973, compared the effectiveness of a protocol-guided nurse to the effective­ Figure 2. A randomized, cross-over study design to compare both process and out­ come. The patients followed the treatment recommendations of the practitioner they ness of a group of physicians with re­ saw last. spect to history taking, physical exam­ ination, simple laboratory observa-

1 8 0 THE JOURNAL OF FAM ILY PRACTICE, VOL. 2, NO. 3,1975 rics, radiology, and surgery. The physi­ tions and management (Figure 2). physician to indicate on a special sheet Women who had either dysuria, fre­ cians knew the study goals and design, a judgment on these protocol-recom­ quency of urination, vaginal discharge, and all inquiries were answered; how­ mended decisions. The physician was or vaginal irritation were randomly al­ ever, the protocol was not reviewed asked whether the protocol recom­ located to one of two groups. with any of these physicians. mendation was reasonable, not wheth­ One group was seen first by a nurse, Patients assigned to the other group er it corresponded exactly to what he an RN without practitioner experi­ were seen first by a physician and then or she had recommended or would rec­ ence, and then by one of 13 partici­ by the nurse. Neither the physician ommend. If the physician was opposed pating physicians. Eight of the 13 nor the nurse knew the other’s find­ to the nurse-protocol disposition or physicians were regular staff members ings. The order of patient contact was therapy, the physician’s decisions pre­ 0f the Student Health Service, and five reversed to control for the patient’s vailed. The information on these pa­ were part-time residents at the UCLA sensitization to the history and physi­ tients was removed from the analysis Hospital in the departments of pediat­ cal examination by the first observer. of outcome results. The nurse examined the urine sedi­ The patients were asked to tele­ ment and the vaginal smear without phone two days later for the results of knowing the results of the Student the urine culture. If the urine culture Health Service laboratory examination was positive the patient was asked to Table 1. Data Collection Concordance of the same specimen. return for a repeat culture one week Between The nurse (on the protocol form) after termination of therapy. Within a Physicians and Nurse-Protocol (N-P)* and the physician (on a similar form week after the clinic visit all patients that did not include the protocol out­ were contacted by telephone and ques­ line) recorded their history and physi­ tioned about the presence, absence, or Physician N-P Identical cal examination findings, and after as­ alleviation of symptoms and about Error Error certaining the laboratory results they complications. committed themselves to a presump­ Each protocol report was reviewed tive diagnosis and plan. The patients by Dr. Greenfield to verify that the seen last by a physician followed the protocol was followed accurately. History 139 6 (4) 1 (0) physician’s recommendations (physi­ Results Physical Exam 137 0 9 (0) cian group). Those patients seen last by the nurse followed the protocol- Five of the 151 patients admitted recommended disposition and therapy to the study did not complete the * ( ) = Resulting in altered diagnosis (nurse-protocol group). The nurse gave study because they refused to be ex­ or management. the physician the protocol recommen­ amined twice or because of an error in dations, for treatment or for referral triage. The data were analyzed for the to a physician, and requested the remaining 146 patients.

Table 2. Diagnosis Concordance Between Nurse-Protocol (N-P) and Physician for 130 Patients*

N-P Diagnosis Physician Diagnosis

UTI Urethral Monilia Nonspecific Trichomonas UTI and Othert Syndrome Vaginitis Vaginitis or Urethritis

Urinary tract infection 28

Urethral syndrome or urethritis 6

Monilia 37

Nonspecific vaginitis 1 41

Trichomonas 1

Urinary tract infection and vaginitis 2 8

Othert 6

*UTI = urinary tract infection. (Includes no pathology found, trichomonas and monilia together, and suspected allergic reactions to vaginal insertions.

181 THE JOURNAL OF FAM ILY PRACTICE, VOL. 2, NO. 3, 1975 There were five patients over 30 the patient would have gone home visit for the same complaint or recent­ years old; the others were 18 to 22 without seeing a physician. The proto­ ly taken medication interfering with years old. The full-time Student col directed that the patient be re­ the diagnosis and therapy, three were Health Service physicians saw 99 pa­ ferred to the doctor in 16(11 percent) for symptoms of generalized toxicity tients. The remaining 47 patients were cases. Concordance of diagnosis in the and the remainder were for miscellane­ seen by the five part-time residents. remaining 130 patients is seen in Table ous reasons. The protocol did not fail Concordance between physicians 2. The diagnoses were made after re­ to recommend referral for any patient and the nurse with respect to the pa­ view of the laboratory tests results; the physicians felt had to be examined tient’s medical history is recorded in agreement was virtually complete. In by a physician. Table 1. Of the 146 histories, 139 two cases a diagnosis of vaginitis was The outcome of symptoms was were essentially identical. Of the seven confirmed by the nurse when the evaluated in the nurse-protocol-treated discrepancies, there were six cases of physician had noted only urinary tract and physician-treated groups. Of 76 error by physicians. In four cases, the infection; in both instances, on review, patients allocated to the nurse treat­ physicians concurred that they had the physician agreed with the nurse- ment group, eight did not receive erred in not pursuing symptoms sug­ protocol diagnosis and therapy. In one nurse-protocol management because of gesting either vaginitis or urinary tract case the laboratory found monilia on a referral (six cases), physician disagree­ infection, and management was altered Gram’s stain of vaginal discharge of a ment (one case), or no identifiable when either condition was found. In woman who had been diagnosed by pathology (one case). Similarly, five two cases, physicians agreed that they the nurse as having nonspecific vagini­ patients of the 70 assigned to the had not acquired a presumptive his­ tis. physician treatment group were ex­ tory of urinary tract problems. The There was virtually complete agree­ cluded from analysis of specific thera­ nurse did not detect a past history of ment on therapy. In one case the py because of referrals to subspecial­ significant urinary tract infection in physician agreed with the nurse-proto­ ists or other complications. The out­ one patient. col diagnosis but preferred another come of symptoms for the remaining Concordance on the results of treatment. In the 129 other cases, 68 patients treated by nurse-protocol physicial examination was similar. Of physicians judged the protocol treat­ and 65 patients treated by physicians 146 cases, 137 were virtually identical. ment plan to be “reasonable.” Physi­ are recorded in Table 3. Of the 65 In nine cases the nurse made an error, cians concurred with the nurse-proto- physician-treated patients all but two based on concurrent physician assess­ col diagnosis and management (specif­ reported alleviation or improvement of ment. These nine errors were not inde­ ic therapy or referral) in 144 of the their symptoms. Similarly, only two of pendently verified. In no case did any 146 cases. the 68 nurse-protocol patients report­ of these physical examination errors Beside advising specific treatment, ed no improvement. The three pa­ result in a different management deci­ the protocol recommended a review tients with unimproved vaginitis were sion: in five cases the nurse did not by a physician in 11 cases, nine for a thought to have nonspecific vaginitis recognize monilia in vaginal discharge, suspected history of urinary tract dis­ on the first visit and were later treated but monilia was shown by laboratory ease and two because the patient had for monilia infection. One patient who tests; in four cases physicians noted recently taken medication potentially had a urinary tract infection (treated costovertebral angle tenderness that interfering with diagnosis and therapy. by the physician) had an allergic reac­ the nurse had not noted, but they did There was agreement on the need for tion to sulfisoxazole. not specifically diagnose pyelonephri­ review of all 11 records. Results of treatment with antibi­ tis or alter management of routine uri­ All 16 patients referred to the otics of urinary tract infection are nary tract infection. physician by the protocol-directed shown in Table 4. Nine of the physi­ Evaluation of the laboratory work nurse were referred appropriately, ac­ cians’ 15 culture-positive patients had done by the nurse indicated that the cording to the physicians. Of the 16 repeat cultures one week after treat­ Health Service’s laboratory findings referrals, seven were for either a return ment terminated. Eight of nine cul- agreed with the nurse’s in 54 of 58 urinalyses. In the four cases of dis­ agreement, the physician who had seen the patient examined the sediment and Table 3. Symptomatic Outcome by Treatment Group and Diagnosis confirmed the nurse’s observation. The laboratory failed to note the presence of monilia in vaginal secretions in nine UTI/US* Vaginitis Both cases of the 39 in which the nurse had detected monilia. In all of these cases, Improved the nurse’s findings were confirmed by Total Improved Total Improved Total the physician. The nurse failed to note monilia on one specimen found to be 3 positive by the laboratory. There was Nurse-protocol 16 16 49 47 3 no independent judgment made on the Physician 23 22 35 34 7 7 specimens reported as negative by both nurse and laboratory. The protocol’s diagnostic accuracy *Urinary tract infection/urethral syndrome. was evaluated in those cases in which

1 82 THE JOURNAL OF FAMILY PRACTICE, V O L . 2, NO. 3,1975 specific skills required for the manage­ Most urinary tract infections are self­ Table 4- Culture Results of Antibiotic ment of the specific complaints we limited diseases in terms of symp­ T r e a t m e n t for Urinary Tract Infection studied. The few “errors” in physical toms.6 The problem with vaginitis is examination made by the nurse were not difficulty in diminishing symp­ not verified by an independent ob­ Positive Total Culture toms but recurrence that is trouble­ Group server, and, even if real, they did not some and chronic, it is accepted, how­ Patients Before Recul­ Sterile prevent her from reaching a proper T reat- tured A fter ever, that any given episode is not dif­ diagnostic conclusion because of the ment 3 Weeks ficult to treat symptomatically. built-in checks of the protocol. As At this point the most significant • with all clinical medicine, decisions are indicator of outcome for urinary tract seldom made on the basis of a single Physician 15 9 8 infections is a negative urine culture branch in logic. after treatment. Although the contro­ 12 10 Nurse-protocol 14 That the nurse did as well in the versy as to whether symptomatic and laboratory determinations as the Stu­ Total 29 21 18 asymptomatic bacteriuria lead to dent Health Service laboratory is to be chronic is unsettled, we expected. The nurse learned a few lab­ adopt the conservative approach that oratory procedures very well and be­ bacteria should be eradicated. Eight of tures were sterile. Similarly, of the 14 came expert at them while the labora­ the nine physicians’ patients and ten positive cultures in the nurse-protocol tory technicians were doing many pro­ of the 12 nurse-protocol patients had patients group, 12 were recultured and cedures, without emotional investment sterile urine cultures at three weeks. ten of these were sterile within three in a given procedure or in any patient. Although the numbers are small, these weeks. This concentration also pertains to the results compare favorably with pre­ As determined by review of the accuracy in following the protocol. vious studies.6,8,9 protocols, very few minor errors and The nurse (S. S.) was doing only this Beside providing care of demon­ no substantive errors were made in fol­ study at the time, and she worked strably high quality, the nurse using lowing the logic. closely with one of the physicians the protocol showed a significant po­ (S.G.). tential for saving physician time. Al­ Discussion The quality of the diagnostic, thera­ most 90 percent of the patients could The study seems to support the peutic, and disposition recommenda­ have been managed completely by the three hypotheses tested: the nurse ac­ tions of the protocol was also good, as nurse. For those patients referred to curately collected the relevant clinical reflected by both process and outcome the physician, a nurse’s completing a data; the protocol made appropriate measurements. Diagnostic impressions preliminary examination would have diagnostic, therapeutic, and disposi­ by the nurse-protocol were in almost saved considerable time, depending on tion recommendations; use of the pro­ complete concordance with those of the complexity of the problems and tocol by the nurse saved physician the physicians. The protocol’s pro­ the physician’s style of practice. time. Thus the protocol met the two visions for a detailed history allowed The study had several possible essential requirements for its wide­ the nurse to make slightly more accu­ shortcomings. One is that the protocol spread use: the quality of care was rate diagnoses than the physicians. was tested with only one nurse. This good, as reflected by both process and Physicians agreed that the nurse-proto- affects solely the question of how well outcome criteria, and use of the proto­ col treatment plan was reasonable in nonphysicians collect clinical data and col introduced the anticipated efficien­ virtually all cases. not the question of how appropriate cy of care. Perhaps most important, considera­ the protocol was in recommending Because urinary tract symptoms, tion of the reasons for physician re­ treatment or referring patients to the vaginitis symptoms, and bacteriuria ferral for 16 patients indicated that physician. We feel that the skills re­ can be self-limited, laboratory cultures the protocol was conservative, as in­ quired for the protocol are so explicit and outcome of symptoms are not suf­ tended, and entirely safe. Those pa­ and circumscribed that a nurse with ficient criteria to validate protocol de­ tients who would have been sent home moderate ability should be able to cisions. It is also necessary to evaluate by the protocol rules without seeing a master them. Further, as indicated ear­ the process of medical care — to com­ physician had no complications as as­ lier, the protocol is so constructed that pare the nurse-protocol with the physi­ certained by our follow-up and by individual data collection errors will cians’ procedure regarding the physician estimation of the com­ have only a small effect on decisions. thoroughness and accuracy of the clin­ plexity of the problem. All potential A second possible shortcoming is ical data collected and the diagnosis, complications were referred to physi­ the restricted age group of the pa­ therapy, and disposition. The study cians, and we anticipate that the pro­ tients. Use of the protocol in older pa­ design permitted independent evalua­ tocol is conservative enough to con­ tients more prone to chronic diseases tion of these aspects of the process. tinue this degree of safety with larger might increase the number of patients The results were favorable regarding numbers of patients. referred to the physician, thus re­ the thoroughness and accuracy of the Study of the outcome of care also ducing the protocol’s efficiency. It is clinical data collected by the nurse. supported these conclusions. Patients unlikely that, when used with a differ­ The concordance between the nurse treated by the protocol-directed nurse ent patient population, the protocol and physicians in the history, physical obtained the same high degree of would be less effective in directing examination, and laboratory data indi­ symptom relief as those treated by the treatment or referral. This particular cates that a nurse can be trained in the physician, which is not surprising. aspect should be investigated further.

1 83 THE JOURNAL OF FAM ILY PRACTICE, VOL. 2, NO. 3, 1975 It can also be argued that bias and audit the performance of various 110-73-335, Bureau of Health Services r among the physicians may have affect­ health care professionals has been re­ search and Evaluation, Health Dp, Administration, Public Health Servicin'* ed their judgments about the proto­ jected by some people as being too partm ent of Health, Education, and Welfare col’s diagnostic, therapeutic, and dis­ constraining, cumbersome, time-con­ position decisions. The physicians suming, and limited, because of the 1. Komaroff AL, Black WL Flatl»„>, were not instructed on the logic be­ diversity of clinical problems, and pro­ et al: Protocols for physician assist!^ hind the protocol decisions. In each ductive of mindless, mechanical, and M a n a g e m e n t o f diabetes and hyperteTs S' N Engl J Med 290:307-312, 1974 case the physicians committed them­ potentially dangerous behavior. We 2. Sox HC Jr, Sox CH, Tompkins Rr selves to a diagnostic, therapeutic, and have discussed some of these concerns The training of physician's assistants Th use o f a c lin ic a l a lg o rith m system for disposition plan before knowing the elsewhere.1,10 Our experience in this care, audit of performance, and educate N Engl J Med 288: 818-824, 1973 0n' nurse-protocol plan. Over time, how­ study and with other protocols does 3. Greenfield S, Bragg FE, McC . ever, the physicians may have made not support these concerns; neverthe­ D L , e t al: A n u p p er respiratory tracuo™ p la in t p ro to c o l fo r physician-extender. proper inferences about the protocol, less, additional carefully evaluated ex­ Arch Intern Med 133:294-299, 1974 WS' which would introduce a bias. This perience is required. 4. Sacks TG, Abramson JH: Screen™ tests fo r b a c te riu ria . A v alid ity study jama bias could have affected the physi­ We conclude that this protocol met 201:79-82, 1967 cians’ agreement with the appropriate­ the criteria for quality and can be used 5. Kunin CM: The quantitative signifi cance of bacteria visualized in the unstained ness of the referrals. Two of us (S. G. by appropriately trained health profes­ urinary sediment. N Engl J Med 265-5R0 5 9 0 , 1 96 1 3 and S. S.) involved in the daily aspects sionals other than physicians to man­ 6 . Brum fitt W, Reeves DS: Recent de­ of the study ascertained from informal age complaints of urinary tract infec­ v e lo p m e n ts in th e tre a tm e n t of urinary tract in fe c tio n . J In fe c t Dis 1 2 0 :6 1 -8 1 , 1969 conversations with the physicians that tions and vaginitis. The protocol saves 7. Brooks D, Maudar A: Pathogenesis they were clearly disposed to accept physician time, allows ease of evalua­ of the urethral syndrome in women and its diagnosis in general practice. Lancet 2 893 the protocol recommendations on re­ tion of medical care, and facilitates 8 9 8 , 1 9 7 2 ferral to a physician when they real­ teaching the management of these 8 . B ru m fitt W , Pursell R: Double-blind tria l to c o m p a re a m p ic illin , cephalexin, cc ized that the protocol had been devel­ complaints without sacrificing high trim oxazole, and trimethoprim in treatment of urinary infection. Br Med J 2:673-676 oped after careful review of the litera­ standards of medical care. 1 9 7 2 ture and analysis by specialist consul­ 9. Hulbert J: Gram-negative urinary in- tants. If the physicians were influ­ fection treated with oral penicillin G. Lan Acknowledgments cet 2:1216-1219, 1972 enced toward higher quality medical 10. Komaroff AL, Reiffen G, Sherman The authors thank Dr. M. M. Osborne, Jr. H: Problem-oriented protocols for physi care, this bias adds to the value of the and Dr. John Champion of the UCLA Stu­ cian-extenders. In Walker HK, Hurst W, protocol rather than detracts from it. dent Health Service for their cooperation Woody N (eds): Applying the Problem- and Dr. Charles E. Lewis for guidance and Oriented System. New York, Medcom The use of protocols to train, guide, advice. Grant support: contract HSM 1973, pp 186-196