Mycobacterium abscessus post-injection abscesses from extrinsic contamination of multiple-dose bottles of normal saline in a rural clinic☆

Case finding 

We defined a case as an abscess or persistent induration at the site of any injection given between June 1, 2006, and February 1, 2007, at any of the four village clinics affiliated with one township hospital. We found cases by searching medical records in all four village clinics, follow-up of patients who had received intramuscular (IM) and/or intravenous (IV) injections at clinic Y, and a house-to-house search of village Y. Physicians at the local township hospital had performed incision and drainage of abscesses of patients at the time of presentation for clinical care. They had submitted samples of pus to the clinical laboratory that served their hospital. We obtained AFB isolates from the clinical laboratory, determined their antibiotic sensitivities, and identified them using standard biochemical techniques.

Retrospective cohort study 

To identify which drug, diluent, or instrument was the source of this outbreak, we identified all orders for IM or IV injections at clinic Y from August to October 2006. From these we determined the patient, medication, diluent, and route of each injection. We then compared rates of abscess incidence for each of these exposures.

Evaluation of injection practices and clinic environment 

We reviewed injection procedures in the treatment room. We ascertained the manufacturer and lot numbers of all injectable drugs and the source of other components of each injection. Using standard environmental sampling methods,9, 10 we took environmental specimens of air, materials, surfaces, and water (including well water supply) to culture for mycobacteria.

We undertook this investigation to respond to an acute problem of adverse events. The principal objective was to identify the mode of exposure and source of infection to terminate the outbreak and benefit the affected community. Responses of this nature are not considered research and accordingly do not require institutional review board approval. Moreover, we performed no additional procedures, tests, data collection, or data analysis above those needed to resolve the immediate public health problem.

Case finding and characteristics 

We identified 35 cases, including 33 in residents of village Y and two in residents of a neighboring village. All 35 had received injections from clinic Y. The onset of first noticeable symptoms was from October 14, 2006 through January 22, 2007. The antecedent injections had been given from August to October 2006. Ten cases occurred in males and 25 in females. The ages of patients ranged from 8 months to 96 years. All 35 had abscesses and 69% (n=24) also had persistent induration at the injection site. Excluding two patients with upper respiratory symptoms, 21% (n=7) of patients also had fever from >37°C to 38.5°C. Twenty-six percent (n=9) had inguinal lymphadenopathy. One case patient had abscesses in both buttocks and the rest had abscesses in either the left (49%; n=17) or right (49%; n=17) buttock. Some case patients reported pain or hyperalgesia at the injection site. Chest films were clear for all case patients. Twenty-eight cases were treated by incision and drainage and one by needle aspiration. Two abscesses drained spontaneously. Four patients did not receive treatment. Pus from eight of 26 abscesses yielded M. abscessus. All eight isolates were resistant to isoniazid, rifampin, p-aminosalicylic acid, levofloxacin, and capreomycin. They were sensitive to clarithromycin, amikacin, and ethambutol.

Retrospective cohort study 

During the period of the cohort analysis, August through October 2006, 838 injections were given to 604 patients at clinic Y. Of these patients 5.8% (n=35) developed a case. All 35 cases occurred in 184 patients (attack rate=19.0%) who had received IM injections with one or more of four different drugs (cefradine, ribostamycin, dexamethasone, and ribavirin) dissolved in NS (Table 1). Two of these drugs, dexamethasone and ribavirin, did not require reconstitution in NS but were sometimes mixed in the same syringe with drugs needing reconstitution in NS (cefradine and ribostamycin). When given in this combination, dexamethasone and ribavirin were associated with an increased risk of abscesses. In contrast, when given without NS, they had zero risk. Among the 184 patients given IM injections with NS, none of the four individual medications was associated with a greater risk of abscesses. Finally, the risk of having an abscess increased as the number of NS-containing injections increased (p<0.05; Chi-square for trend=6.1). No patient receiving an injection without NS developed a case.

Table 1. Exposures of 35 abscess case patients among 604 patients exposed to 838 injections from clinic Y by drug and route of injection; Guangdong Province, China, August–October 2006

	Injection type and druga	Exposed		Unexposed		% Exposed	Attack rate (%)		RR	95% CI
		Case	Total	Case	Total		Exposed	Unexposed
	Injection route
	Any IM	35	527	0	77	87	6.6	0	∞	-
	Any IV	5	153	30	451	25	3.3	6.7	0.49	0.19–1.2
	Only IV	0	77	35	527	13	0	6.6	0	-
	IM injections
	Any with NS	35	184	0	420	30	19.0	0	∞	-
	Any without NS	16	422	19	182	70	3.8	10.4	0.36	0.19–0.69
	Only without NS	0	343	35	261	57	0	13.4	0	-
	IM with NS in syringe by drugb
	Ribostamycin	16	67	19	117	36	23.8	16.2	1.5	0.81–2.7
	Cefradine	24	129	11	55	70	18.6	20	0.93	0.49–1.8
	Ribavirin	8	45	27	139	24	17.8	19.4	0.92	0.45–1.9
	Dexamethasone	26	140	9	44	76	18.6	20.5	0.91	0.46–1.8
	IM without NS in syringe by drug
	Dexamethasone	0	82	35	522	14	0	6.7	0	-
	Ribavirin	0	7	35	597	1.2	0	5.9	0	-

RR, risk ratio; CI, confidence interval; IM, intramuscular injection; IV, intravenous injection; NS, normal saline.

a Patients may have been exposed to more than one injection type and drug. b N=184 patients.

To better define the period of exposure we calculated a case-injection/total injection rate in intervals of one-third month. We also identified 18 case patients who had received only one intramuscular injection. The case-injection/total injection rate rose sharply from a baseline during the middle of August (Figure 1). All cases had had one or more IM injections containing NS between August and October 2006. Injections of case patients who had received only one injection began to appear in mid-August. The clinic began to use 100ml bottles of NS in June when it ran out of 2ml vials of NS. Injections with NS were stopped on October 23, 2006. Using the nine cases with single injections who could report an onset day, we estimated the median incubation period to be 31 days (range 13–87 days).

View Large Image Download to PowerPoint

Figure 1. Intramuscular injections with normal saline by date, with rate for all cases and count for cases with only one injection; clinic Y, Guangdong Province, China, August–October 2006.

Assessment of injection practices and clinic environment 

Beginning in June 2006, the nurse responsible for giving injections at the clinic began using multiple-dose bottles (100ml) of NS (Figure 2) to dilute cefradine and ribostamycin. The nurse routinely left a 16-gauge needle inserted through the septum of the bottle. When the bottle was not in use the nurse attached a syringe to the hub of this needle. To prepare an injection the nurse would detach the syringe from this needle, lay the syringe on the table and attach another syringe to be used for injection to the needle and withdrew NS. She would use this second syringe and NS to dilute the drug. She would then aspirate drugs into the second syringe, give the injection, and reattach the first syringe to the needle in the saline bottle. The nurse stated that she changed the syringe used to plug the needle hub in the NS bottle whenever she began using a new bottle of NS, about every three to seven days. She never used disinfectant to clean the septum of the NS bottle. To prepare the skin for injection the nurse took cotton balls directly from original bags and wet them with 75% ethanol. The clinic stopped using 100ml NS for IM injections on October 23, 2006 and stopped cefradine or ribostamycin injections on November 13, 2006. The nurse used the same lot of NS from June through October.

View Large Image Download to PowerPoint

Figure 2. Normal saline bottle showing the open 16-gauge needle and syringe used to plug the needle hub.

Environmental samples taken in December revealed M. abscessus from the clinic well water. The well water was not chlorinated or otherwise treated or disinfected. We found AFB from the top of the cabinet for storing syringes, and the sewage from the clinic, but these were not M. abscessus. Sampling of 37 other sites including the injection table, the cloth used to wipe down the table, the sink and drain, the antiseptic solution, air, syringes, and needles did not yield AFB. The nurse's hands were negative for AFB and she had no pulmonary disease.

Two of the three other clinics served by the township hospital used NS from the same lot number and from the same manufacturer. From August to December 2006 these two clinics attended to 2649 outpatients and used 240 bottles of the same lot of NS to prepare injections. The nurses in these two clinics used the same technique for preparing drugs for injection as the nurse at clinic Y with one difference. They opened a new bottle of NS each day. We identified no post-injection abscesses among patients of those two clinics.

Village Y had a population of 3887 registered residents. Using this population, we can estimate from our cohort data that 14% (n=527) received IM and 4% (n=153) received IV drugs over the three months covered by our investigation. The clinic physician reported that he used parenteral drugs to treat minor illnesses as he referred patients with more severe illnesses directly to the township hospital.